dementia research paper

Diagnosis and Management of Dementia: Review

Affiliations.

• 1 Rush Alzheimer's Disease Center, Rush University Medical Center, Chicago, Illinois.
• 2 Department of Neurological Sciences, Rush University Medical Center, Chicago, Illinois.
• 3 Department of Family Medicine, Rush University Medical Center, Chicago, Illinois.
• PMID: 31638686
• PMCID: PMC7462122
• DOI: 10.1001/jama.2019.4782

Importance: Worldwide, 47 million people live with dementia and, by 2050, the number is expected to increase to 131 million.

Observations: Dementia is an acquired loss of cognition in multiple cognitive domains sufficiently severe to affect social or occupational function. In the United States, Alzheimer disease, one cause of dementia, affects 5.8 million people. Dementia is commonly associated with more than 1 neuropathology, usually Alzheimer disease with cerebrovascular pathology. Diagnosing dementia requires a history evaluating for cognitive decline and impairment in daily activities, with corroboration from a close friend or family member, in addition to a thorough mental status examination by a clinician to delineate impairments in memory, language, attention, visuospatial cognition such as spatial orientation, executive function, and mood. Brief cognitive impairment screening questionnaires can assist in initiating and organizing the cognitive assessment. However, if the assessment is inconclusive (eg, symptoms present, but normal examination findings), neuropsychological testing can help determine whether dementia is present. Physical examination may help identify the etiology of dementia. For example, focal neurologic abnormalities suggest stroke. Brain neuroimaging may demonstrate structural changes including, but not limited to, focal atrophy, infarcts, and tumor, that may not be identified on physical examination. Additional evaluation with cerebrospinal fluid assays or genetic testing may be considered in atypical dementia cases, such as age of onset younger than 65 years, rapid symptom onset, and/or impairment in multiple cognitive domains but not episodic memory. For treatment, patients may benefit from nonpharmacologic approaches, including cognitively engaging activities such as reading, physical exercise such as walking, and socialization such as family gatherings. Pharmacologic approaches can provide modest symptomatic relief. For Alzheimer disease, this includes an acetylcholinesterase inhibitor such as donepezil for mild to severe dementia, and memantine (used alone or as an add-on therapy) for moderate to severe dementia. Rivastigmine can be used to treat symptomatic Parkinson disease dementia.

Conclusions and relevance: Alzheimer disease currently affects 5.8 million persons in the United States and is a common cause of dementia, which is usually accompanied by other neuropathology, often cerebrovascular disease such as brain infarcts. Causes of dementia can be diagnosed by medical history, cognitive and physical examination, laboratory testing, and brain imaging. Management should include both nonpharmacologic and pharmacologic approaches, although efficacy of available treatments remains limited.

Publication types

• Research Support, N.I.H., Extramural
• Research Support, Non-U.S. Gov't
• Alzheimer Disease / diagnosis
• Alzheimer Disease / therapy
• Cholinesterase Inhibitors / adverse effects
• Cholinesterase Inhibitors / therapeutic use
• Dementia / diagnosis*
• Dementia / therapy*
• Excitatory Amino Acid Antagonists / adverse effects
• Excitatory Amino Acid Antagonists / therapeutic use
• Memantine / adverse effects
• Memantine / therapeutic use
• Neuroimaging
• Neuropsychological Tests
• Cholinesterase Inhibitors
• Excitatory Amino Acid Antagonists

Grants and funding

• P30 AG010161/AG/NIA NIH HHS/United States
• R01 AG040039/AG/NIA NIH HHS/United States
• R01 NS084965/NS/NINDS NIH HHS/United States
• RF1 AG059621/AG/NIA NIH HHS/United States

• - Google Chrome

Intended for healthcare professionals

• Access provided by Google Indexer
• My email alerts
• BMA member login
• Username * Password * Forgot your log in details? Need to activate BMA Member Log In Log in via OpenAthens Log in via your institution

Search form

• Advanced search
• Search responses
• Search blogs
• Dementia: timely...

Dementia: timely diagnosis and early intervention

• Related content
• Peer review
• Louise Robinson , general practitioner and professor of primary care 1 ,
• Eugene Tang , NIHR academic clinical fellow in general practice 1 ,
• John-Paul Taylor , senior clinical lecturer and honorary consultant in old age psychiatry 2
• 1 Institute of Health and Society, Newcastle University, Newcastle upon Tyne, NE2 4AX, UK
• 2 Institute of Neuroscience, Newcastle University, Newcastle upon Tyne, UK
• Correspondence to L Robinson a.l.robinson{at}ncl.ac.uk

The bottom line

Dementia is a major global health problem; in the absence of a cure there is increasing focus on risk reduction, timely diagnosis, and early intervention

Primary and secondary care doctors play complementary roles in dementia diagnosis; differential diagnoses include cognitive impairment due to normal ageing and depression

Cost effective drug (acetylcholinesterase inhibitors) and non-drug interventions such as cognitive stimulation therapy exist that help to delay cognitive deterioration and improve quality of life; information provision and practical support are also important

Discussions about a person’s wishes for future care should occur at an early stage of illness while the person has mental capacity

Family carers of people with dementia are at high risk of physical and mental illness as a consequence of caring and they require equal attention and support

Dementia describes a clinical syndrome that encompasses difficulties in memory, language, and behaviour that leads to impairments in activities of daily living. Alzheimer’s disease is the most common subtype of dementia, followed by vascular dementia, mixed dementia, and dementia with Lewy bodies. Because the global population is rapidly ageing, dementia has become a concern worldwide 1 ; the illness places considerable burden on individuals and their families and also on health and social care provision.

Sources and selection criteria

We searched for articles through Medline, PubMed, and the Cochrane database of systematic reviews from January 2006 to December 2014—the period after publication of the current UK national dementia guidance 9 —using the search terms “dementia”, “Alzheimer’s”, “carer”, and “caregiver”. Additional searches were carried out for specific subsections—for example, “pharmacologic treatment” and “non-pharmacologic interventions/strategies/treatment”. Where possible, we focused on systematic reviews, meta-analyses, and high quality randomised controlled trials. We included only articles in English and excluded those published in non-peer reviewed journals. Recommendations in this review are derived from the most recent international and UK national guidance 9 on evidence based practice in dementia care and the authors’ interpretation of the included evidence.

By 2050 an estimated 135 million people worldwide will have dementia. In 2010 the global cost of dementia care was estimated at $604bn (£396bn; €548bn) and estimated to increase to $1tr by 2030. 1 Of all chronic diseases, dementia is one of the most important contributors to dependence and disability. 2 3 In the absence of a cure, a professional belief that nothing can be done has contributed to delays in diagnosis. 4 However, increasing evidence showing that dementia may be preventable 1 5 has led to an international focus on earlier diagnosis and intervention. 6 This review aims to summarise current evidence and best practice in the diagnosis and early intervention in dementia care.

Patient and public involvement

Patient and public involvement in this clinical review has been achieved through several processes: the inclusion of patients and the public in the groups responsible for developing the national guidelines referenced in this review; liaising with patient and public representatives from the National Institute of Health Research Dementia and Neurodegenerative Diseases Research Network who contributed to systematic reviews included in this review 4 9 ; and asking the UK Alzheimer’s Society to comment on the final draft of the paper and provide up to date information resources for patients and carers.

Why is timely diagnosis important?

In some countries the introduction of a national dementia strategy has led to greater emphasis on earlier diagnosis, although population based screening is not recommended as dementia does not fulfil the criteria of a condition suitable for screening. 7 With evidence from large longitudinal cohort studies showing that the prevalence of dementia is declining globally, there is now greater emphasis on prevention and risk reduction. 1 5 In England, policy has rightly or wrongly influenced the introduction of case finding in high risk groups—including people over 75 years of age, as age is the strongest risk factor for dementia—and those with high vascular risk, Parkinson’s disease, and learning disabilities. 8 The policy comprises proactive memory assessment of people in both primary care and acute hospital settings who may not have symptoms; however, there is little evidence that such initiatives, which inevitably lead to increased referrals to specialist services, are cost effective and whether they are distressing to patients. 4 6

How can clinicians recognise dementia?

Diagnosing dementia can be difficult owing to its insidious onset, symptoms resembling “normal ageing” memory loss, and a diversity of other presenting symptoms—for example, difficulty in finding words or making decisions. 10 An individual’s ability to accommodate, compensate, or even deny his or her symptoms in the early stages should also be considered. The individual’s family may also have noticed difficulties in communication and personality or mood changes; family concern is of particular importance. 9 Increasing frequency of patients’ visits to their general practice, missed appointments, or confusion over drugs may also be warning signs. 8

Diagnosis of subtype is important given differences in management, disease course, and outcomes for different dementias; awareness of early symptoms in less common dementias can assist generalists in deciding to which specialist services patients are referred (box 1). Duration over which symptoms have developed is also important, with Alzheimer’s disease tending to have a more insidious onset than vascular dementia.

Box 1 Examples of less common dementias and their early presenting symptoms

Vascular dementia.

Wide range of signs and symptoms depending on extent, location, and severity of the cerebrovascular disease

Symptoms can develop abruptly after a stroke or more insidiously with small vessel disease

Memory loss can be a feature but typically is less noticeable than in Alzheimer’s disease. Language, information processing, decision making, and visuospatial deficits can also be found

Mood changes and apathy are common symptoms; can co-occur with Alzheimer’s disease and this is termed mixed dementia

Frontotemporal dementias

More common in younger age groups (50-60 years)

The most common clinical type is behavioural variant frontotemporal dementia, with changes in personality and behaviour. Disinhibition and impulsiveness can be features. Memory function is typically intact early on

Dementia with Lewy bodies

Complex visual hallucinations are a key feature. In the early stages they may only occur during periods of physical stress (for example, infections) or at night time and may be followed by more subtle visuoperceptual symptoms—for example, illusions

Parkinsonism (tremor, slowed movements, postural instability, shuffling gait) is also a feature. Tremor may be less evident, but people with early dementia with Lewy bodies may be slower in movements and more prone to falls

Fluctuations or noticeable variations in cognitive function can occur and can be difficult to separate from delirium

Autonomic symptoms may occur—for example, postural hypotension

Sleep disturbances such as rapid eye movement sleep behaviour disorder (shouting out or moving while asleep) can occur many years before the onset of dementia

Parkinson’s disease with dementia

As many as 80% of patients with Parkinson’s develop dementia

Symptoms are similar to those of dementia with Lewy bodies, although motor Parkinson’s symptoms typically predate cognitive and psychiatric symptoms by more than a year

Posterior cortical atrophy

A less common form of Alzheimer’s disease, which tends to affect younger people (50s and 60s)

Visual agnosias (difficulties with recognising faces, objects, or perceiving more than one object at a time), apraxias (motor planning difficulties), acalculia (difficulty with calculation), and alexia (difficulty reading) are symptoms

Memory typically preserved early on

Other uncommon to rare causes of dementia

Alcohol related dementia, Creutzfeldt-Jakob disease, HIV related cognitive impairment, Huntington’s chorea, corticobasal syndrome, movement related dementias (for example, progressive supranuclear palsy), multiple sclerosis, Niemann-Pick disease type C, normal pressure hydrocephalus

How is dementia diagnosed?

The role of primary care.

General practitioners are often the first point of contact for patients who are worried that they may have dementia. The role of primary care is to exclude a potentially treatable illness or reversible cause of the “dementia”—for example, depression, vitamin B 12 deficiency, or thyroid disturbance; refer for specialist assessment, especially those with unusual symptoms (neurological, psychiatric, or behavioural changes) or those with major risk factors (for example, important medical comorbidities, psychosocial problems, harm to self); and ensure patients who have mild cognitive impairment (objective cognitive loss not affecting function and daily living activities) are followed up in primary care, and, if their symptoms become more severe, re-referred for specialist assessment.

Initial assessment should include a careful history from both the patient and the main carer, with particular emphasis on disturbance of cognitive function and activities of daily living. A physical examination should be undertaken to look for any focal neurological signs and exclude any visual or auditory problems. Baseline investigations and a brief cognitive assessment, using one of the many tools available (box 2), should also be carried out before referral to secondary care. 9

Box 2 Investigations and brief cognitive assessment tools for dementia in primary care

Blood tests.

Blood tests that should be ordered are: full blood count, erythrocyte sedimentation rate, urea and electrolytes, thyroid function, vitamin B 12 , and folate. Midstream specimen of urine, chest radiography, and electrocardiography may also be needed where clinically appropriate

Brief cognitive assessment tools

General practitioner assessment of cognition 11.

Takes no longer than five minutes to administer and comprises two components: a six item cognitive assessment with the patient and an informant questionnaire (if the cognitive assessment score is equivocal: 5-8 inclusive). Scores >8 are deemed to represent cognitive impairment and <5 intact cognition. Sensitivity 82-85%; specificity 83-86% 12

6 item cognitive impairment test 13

Takes 3-4 minutes to perform and consists of six questions on orientation and memory, although the test may be susceptible to influences of language and education. Scores of 0-7 are considered normal and ≥8 suggest cognitive impairment. Sensitivity 78.5-83%; specificity 77-100% 12

Mini-cog assessment instrument 14

Takes 2-4 minutes to complete and consists of two components, a three item recall and the clock drawing test. Cognitive impairment is considered to be present if people are unable to recall any of the three items or if they recall only one or two items and draw an abnormal clock. Sensitivity 76-99%; specificity 89-96% 12

Memory impairment screen 15

Takes around four minutes to complete and is a brief four item delayed free recall and cued recall memory impairment test. A score of ≤4 indicates possible dementia. Sensitivity 74-86%; specificity 96-97% 12

The mini-mental state examination 16 has traditionally been recommended as the brief cognitive assessment tool of choice, although copyright restrictions are influencing its use in practice. The tools listed in box 2 have been found to be as clinically and psychometrically robust as the mini-mental state examination 17 ; a clock drawing test may be added to the assessment if it is not already incorporated into the tool. 18 The Addenbrooke’s cognitive examination, 19 especially the revised version, has superior diagnostic accuracy to the mini-mental state examination but takes about 25 minutes to complete and has better accuracy in moderate to high prevalence settings. 20 No one brief cognitive assessment tool is more accurate than another and all are inadequate for assessing early or subtle changes, with scores affected by factors such as education. Mini-mental state examination scores are used to indicate the severity of Alzheimer’s disease: mild, scores 21-26; moderate, scores 10-20; moderately severe, scores 10-14; severe, scores less than 10.

Depression masquerading as dementia is probably the most common differential diagnosis and should always be considered; however, they can coexist and depression may precede dementia. If suspected, a trial of antidepressants may be indicated, with reassessment of the individual’s capabilities and cognitive function 6-8 weeks later.

The role of secondary care

Primary care is increasingly taking on a greater role in both the assessment and the long term care of people with dementia; one multicentre randomised controlled trial found no evidence that specialist memory clinics were more effective than general practice services in providing post-diagnostic support. 21 Secondary services have an important role in defining the dementia subtype, dealing with more complex cases, and stratifying which patients with mild cognitive impairment are at greatest risk of developing dementia and most in need of follow-up.

What are the roles of imaging and other investigations?

Imaging, in particular structural scanning (computed tomography or magnetic resonance imaging), is recommended as part of the investigations of people with suspected dementia in UK, 9 European, 22 and US guidelines. 23 Imaging is now also embedded in several modern diagnostic criteria for different dementias, including Alzheimer’s disease and dementia with Lewy bodies. 24 25 26 In modern dementia imaging there is now less focus on “excluding” reversible causes of dementia (for example, tumours) and more on determination of subtype. Structural imaging, particularly magnetic resonance imaging, can also help clarify whether a vascular disease is contributing to the cognitive impairment and thus whether strict adherence to treatment guidance for vascular risks is warranted.

In the United Kingdom, functional neuroimaging, including hexamethylpropyleneamine oxime (HMPAO) single photon emission computed tomography (SPECT) and [18F]fluoro-2-deoxy-D-glucose positron emission tomography (FDG-PET), is available but usually used as a second line approach to assist with subtype diagnoses, particularly where the diagnosis is in doubt. Dopaminergic iodine-123-radiolabelled 2β-carbomethoxy-3β-(4-iodophenyl)-N-(3-fluoropropyl) nortropane (FP-CIT) SPECT imaging is licensed in Europe and in a number of other countries for the diagnosis of dementia with Lewy bodies and may also be helpful where the clinical diagnosis of dementia with Lewy bodies is not clear. 27

What new investigations are emerging in the diagnosis of dementia?

Cerebrospinal fluid sampling is used to exclude inflammatory, infective, and malignancy related causes of dementia and is typically recommended in individuals with rapid cognitive decline, unusual or neurological presentations, or cognitive impairment at less than 55 years of age. 28 More recently there has been a focus on developing cerebrospinal fluid based markers, such as β amyloid and tau, for changes in Alzheimer’s disease that can predate the onset of the dementia, the so called prodromal phase of Alzheimer’s disease. Although such markers have been incorporated into recent diagnostic criteria for Alzheimer’s disease, 25 26 whether they are effective at predicting those who will develop dementia 29 and, more importantly, practically acceptable, makes their widespread clinical use challenging at present.

It is now possible to directly image amyloid in the brain using several positron emission tomography radiotracers, and this imaging technique may have a future role clinically in predicting which people with mild cognitive impairment will develop Alzheimer’s disease. However there is still major heterogeneity in how these scans are interpreted. For example, a recent meta-analysis found that although amyloid imaging has high sensitivity (83-100%) in detecting people with mild cognitive impairment who convert to Alzheimer’s disease related dementia, diagnostic specificities varied considerably between studies (46-88%). 30

What constitutes best practice in early intervention?

Discussing the diagnosis: saying the “d” word.

Health professionals can be reluctant to speak openly and honestly with patients and their families about dementia, with some refraining from using the “D” word. 28 Although initially discussing the diagnosis may be distressing, evidence suggests most people prefer to know if they have dementia in order to access appropriate support and treatment and to plan for the future. 4 31

What options are available after diagnosis?

Drug interventions.

Clinically and cost effective drugs for dementia are available; the emphasis is to improve or maintain function after neuronal damage rather than to alter the underlying pathogenesis leading to the dementia syndrome. Two classes of drugs are currently recommended for symptomatic (Alzheimer’s disease and mixed) dementia 6 32 : acetylcholinesterase inhibitors donepezil, galantamine, and rivastigmine, and N-methyl-D-aspartic acid receptor antagonists such as memantine. At present, acetylcholinesterase inhibitors are the only recommended options to manage mild to moderate Alzheimer’s disease and there is no evidence that one is more efficacious than another 33 ; notwithstanding, a large randomised controlled trial has recently shown that continued treatment with donepezil is associated with cognitive benefits in moderate to severe dementia. 34 Memantine has been approved for people with moderate to severe Alzheimer’s disease or those with intolerance to acetylcholinesterase inhibitors; it has also been used in mild Alzheimer’s disease but the evidence for this is currently lacking despite its frequent off-label use. 35

Non-drug approaches

The evidence base is steadily increasing for non-drug interventions in dementia care, although further research in many areas is still needed. 6 In a large systematic review evaluating both drug and non-drug interventions in dementia care, cognitive stimulation therapy was found to be as clinically and cost effective as the acetylcholinesterase inhibitors 36 ; reminiscence therapy is also recommended in national guidelines. 9 However, the evidence base for innovative service provision such as case management, whereby a case manager, usually a nurse or social worker acts as the main care coordinator between key stakeholders, including primary and secondary care, is mixed. 6 36 Although the evidence base for cost effectiveness is low, 37 specially developed assistive technology—any device or system that allows an individual to perform tasks that they would otherwise be unable to do, or increases the ease and safety with which the task can be performed—to help people with dementia is available and can be useful in relieving carer anxiety and helping people with dementia to remain living at home ( www.atdementia.org.uk/ ).

Information provision

People with dementia and their families require emotional and practical support to help them live as good a quality of life as they can; the family doctor is in a key position to provide ongoing support and advice once the diagnosis is confirmed. 4 6 Voluntary organisations such as Alzheimer’s International provide a wide range of information resources and practical support for people living with all types of dementia ( www.alz.co.uk/ ). Signposting to local sources of support as well as social services and respite care are integral to the consultation. Listening to an individual patient’s difficulties and concerns and providing simple cognitive and emotional strategies in the primary care consultation are beneficial to both patients and their families.

Discussing the future

One important area to be discussed in the earlier stages of dementia, while people still have mental capacity, is personal wishes for future care and also who should make decisions when the patients are no longer able to do so. In dementia, such discussions—termed advance care planning—have been shown to reduce inappropriate hospital admissions towards the end of life, but the evidence base is weak. 38 39 Discussions about advance care planning require both sensitivity and honesty; general practitioners or hospital specialists are well placed to undertake these discussions if they have an established relationship with the patient. After such conversations, patients can formally record their wishes in several ways, including the completion of an advance directive, or “living will” as it was previously known (box 3).

Box 3 Outcomes of advance care planning discussions: international and national terminology

Statement of wishes and preferences —this documents an individual’s wishes for future care and is not legally binding; in the UK this is known as an advance statement

An advance directive for refusal of treatment (or “living will” )—this is a statement of an individual’s refusal to receive specific medical treatment in a predefined future situation. It is legally binding and comes into effect when a person loses mental capacity. In the UK, this is known as an advance decision to refuse treatment

A proxy decision maker or power of attorney —This is a legally binding document whereby an individual (“donor”) nominates another (“attorney”) to make decisions on his or her behalf should he or she lose capacity. In England, following the Mental Capacity Act, this is now known as a lasting power of attorney and there are two separate aspects to lasting power of attorney, one for an individual’s health and welfare and a second for property and financial affairs

Primary care doctors may find it difficult to assess the mental capacity of an individual with dementia; mental capacity may fluctuate with time and also with acute illness. In England, the introduction of the Mental Capacity Act in 2005 provided much needed guidance for health and social care professionals on how to undertake an assessment of capacity and to make decisions in the best interests of adults who lack the mental capacity to do so for themselves (box 4).

Box 4 Assessment of mental capacity (as derived from UK Mental Capacity Act 2005)

Two stage test for determining whether an individual has mental capacity to make a specific decision.

1. Does the patient have an impairment or disturbance of function of the brain?

2. Regarding a specific decision, can the patient:

understand the decision to be made?

retain sufficient information to make an informed decision?

use information appropriately?

communicate their decision?

Practical tips for assessment of capacity:

Information may need to be provided in different forms

General practitioners may need to assess patients on several occasions—that is, if morning is the best time for them

Record information and the two stages described above accurately in patient notes

Refer to experts (old age psychiatry) if in doubt

Caring for family carers

In the UK, two thirds of people with dementia live independently in the community, with most of their care and support provided by family and friends. Such informal carers are more likely to experience depressed mood, to report a higher care “burden,” and to have worse physical health than carers of people with other long term conditions. 40 They may grieve as their family member loses functional and cognitive abilities, and as companionship, affection, and intimacy are affected; this is termed a living bereavement. Notwithstanding the satisfaction carers experience from caring, the support they receive and their ability to seek help when needed influence how they cope. Supporting informal carers, monitoring their health and wellbeing, and providing or referring them for additional practical and psychological support is another crucial role for general practitioners and community care services. 41

Tips for non-specialists

Occasional memory lapses are common as people get older, especially in the presence of stress, depression, and acute physical illness; review the patient after appropriate treatment has been given or a reasonable length of time has elapsed

If you suspect dementia, take a history from both the patient and the main family carer; the latter’s suspicions are often correct

Be aware that certain groups of people are at greater risk of developing dementia—for example, those who have had a stroke and those with Parkinson’s disease

Early identification of modifiable risk factors for dementia may reduce the numbers of people developing dementia in later life

Effective and useful treatments exist for people with dementia; have a low threshold for referring someone with suspicious symptoms for a specialist memory assessment

Assess both the physical and the mental health of the main family carer; supporting informal carers is an important part of dementia care

Additional educational resources

Resources for healthcare professionals.

Alzheimer’s Society. Assessing cognition in older people ( www.alzheimers.org.uk/cognitiveassessment )—a practical toolkit for health professionals

BMJ Group resources: BMJ Learning modules. ( http://learning.bmj.com/learning/module-intro/dementia-primary-care )—describes the management of dementia in primary care

BMJ Quality ( http://quality.bmj.com )—four e-projects to improve quality of care in the areas of support for carers, antipsychotic drug prescribing, timely diagnosis, and palliative care

Resources for patients and carers

Social Care Institute for Excellence. Dementia gateway ( www/scie/org.uk/dementia )—web based information and e-learning resources written by experts mainly for professional carers and supporters

Alzheimer’s Society. The dementia guide: living well after diagnosis ( www.alzheimers.org.uk/dementiaguide or request copies at [email protected] )—comprehensive practical information for people with dementia and families with a recent diagnosis. Includes a free booklet, video case studies, and downloadable translations

Lewy body Society ( http://lewybody.org/aboutdlb )—website of the only charity in Europe exclusively concerned with dementia with Lewy bodies

FTD Talk ( www.ftd.org )—accessible updates and web information for people with frontotemporal dementia from researchers

Alzheimer’s Disease International. Help for caregivers ( www.alz.co.uk/ADI-publications )—a downloadable booklet produced in collaboration with the World Health Organization: practical tips on caring for someone with dementia

Carers UK Factsheets ( http://carersuk.org )—practical information for carers about topics such as benefits and getting help and support

at dementia ( www.atdementia.org.uk )—a website providing information on assistive technology for people with dementia

Cite this as: BMJ 2015;350:h3029

We thank Tim Beanland, knowledge services manager at the Alzheimer’s Society, London for advice. LR is supported by a National Institute for Health Research professorship (NIHR-RP-011-043).

Contributors: LR drafted the outline and overview of the article; all authors contributed equally to the content. LR is the guarantor of the paper.

Competing interests: We have read and understood the BMJ policy on declaration of interests and declare the following interests: none.

Provenance and peer review: Commissioned; externally peer reviewed.

This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/ .

• ↵ Prince M, Albanese E, Guerchet M, et al. World Alzheimer report 2014. Dementia and risk reduction: an analysis of protective and modifiable risk factors. Alzheimer’s Disease International, 2014.
• ↵ Prince M, Knapp M, Guerchet M, et al. Dementia UK. 2nd ed. Alzheimer’s Society, 2014.
• ↵ Prince M, Prina M, Guerchet M. World Alzheimer report 2013. An analysis of long term care for dementia. Alzheimer’s Disease International, 2013.
• ↵ Iliffe S, Robinson L, Brayne C, et al. Primary care and dementia: 1. diagnosis, screening and disclosure. Int J Geriatr Psychiatry 2009 ; 24 : 895 -901. OpenUrl CrossRef PubMed Web of Science
• ↵ Matthews FE, Arthur A, Barnes LE, et al. A two-decade comparison of prevalence of dementia in individuals aged 65 years and older from three geographical areas of England: results of the Cognitive Function and Ageing Study I and II. Lancet 2013 ; 382 : 1405 -12. OpenUrl CrossRef PubMed Web of Science
• ↵ Prince M, Bryce R, Ferri C. World Alzheimer report 2011: the benefits of early diagnosis and intervention. Alzheimer’s Disease International, 2011
• ↵ Alzheimer Europe. National Dementia Strategies (diagnosis, treatment and research): country comparisons. 2013. www.alzheimer-europe.org/Policy-in-Practice2/Country-comparisons/National-Dementia-Strategies-diagnosis-treatment-and-research .
• ↵ Bamford C, Eccles M, Steen N, et al. Can primary care record review facilitate earlier diagnosis of dementia? Fam Pract 2007 ; 24 : 108 -16. OpenUrl Abstract / FREE Full Text
• ↵ National Institute for Health and Care Excellence/Social Care Institute of Excellence (NICE/SCIE). Dementia: supporting people with dementia and their carers in health and social care. (Clinical guideline CG42) 2006. http://guidance.nice.org.uk/CG42 .
• ↵ Kostopoulou O, Delaney BC, Munro CW. Diagnostic difficulty and error in primary care—a systematic review. Fam Pract 2008 ; 25 : 400 -13. OpenUrl Abstract / FREE Full Text
• ↵ Brodaty H, Pond D, Kemp NM, et al. The GPCOG: a new screening test for dementia designed for general practice. J Am Geriatr Soc 2002 ; 50 : 530 -4. OpenUrl CrossRef PubMed Web of Science
• ↵ Yokomizo JE, Simon SS, Bottino CM. Cognitive screening for dementia in primary care: a systematic review. Int Psychogeriatr 2014 ; 26 : 1783 -804. OpenUrl CrossRef PubMed
• ↵ Brooke P, Bullock R. Validation of a 6 item cognitive impairment test with a view to primary care usage. Int J Geriatr Psychiatry 1999 ; 14 : 936 -40. OpenUrl CrossRef PubMed Web of Science
• ↵ Borson S, Scanlan J, Brush M, et al. The mini-cog: a cognitive ‘vital signs’ measure for dementia screening in multi-lingual elderly. Int J Geriatr Psychiatry 2000 ; 15 : 1021 -7. OpenUrl CrossRef PubMed Web of Science
• ↵ Buschke H, Kuslansky G, Katz M, et al. Screening for dementia with the memory impairment screen. Neurology 1999 ; 52 : 231 -8. OpenUrl CrossRef
• ↵ Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975 ; 12 : 189 -98. OpenUrl CrossRef PubMed Web of Science
• ↵ Milne A, Culverwell A, Guss R, et al. Screening for dementia in primary care: a review of the use, efficacy and quality of measures. Int Psychogeriatr 2008 ; 20 : 911 -26. OpenUrl PubMed Web of Science
• ↵ Shulman KI. Clock-drawing: is it the ideal cognitive screening test? Int J Geriatr Psychiatry 2000 ; 15 : 548 -61. OpenUrl CrossRef PubMed Web of Science
• ↵ Mioshi E, Dawson K, Mitchell J, et al. The Addenbrooke’s Cognitive Examination Revised (ACE-R): a brief cognitive test battery for dementia screening. Int J Geriat Psychiatry 2006 :21:1078-85.
• ↵ Larner AJM. A meta-analysis of the accuracy of the Addenbrooke’s Cognitive Examination (ACE) and the Addenbrooke’s Cognitive Examination-Revised (ACE-R) in the detection of dementia. Int Psychogeriatr 2014 :26:555-63.
• ↵ Meeuwsen EJ, Melis RJ, Van der Aa GC, et al. Effectiveness of dementia follow-up care by memory clinics or general practitioners: randomised controlled trial. BMJ 2012 ; 344 : e3086 . OpenUrl Abstract / FREE Full Text
• ↵ Hort J, O’Brien JT, Gainotti G, et al. EFNS guidelines for the diagnosis and management of Alzheimer’s disease. Eur J Neurol 2010 ; 17 : 1236 -48. OpenUrl CrossRef PubMed Web of Science
• ↵ .Jack CR Jr, Albert MS, Knopman DS, et al. Introduction to the recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 2011 ; 7 : 257 -62. OpenUrl CrossRef PubMed Web of Science
• ↵ McKeith IG, Dickson DW, Lowe J, et al. Diagnosis and management of dementia with Lewy bodies: third report of the DLB Consortium. Neurology 2005 ; 65 : 1863 -72. OpenUrl CrossRef PubMed
• ↵ McKhann GM, Knopman DS, Chertkow H, et al. The diagnosis of dementia due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 2011 ; 7 : 263 -9. OpenUrl CrossRef PubMed Web of Science
• ↵ Dubois B, Feldman HH, Jacova C, et al. Advancing research diagnostic criteria for Alzheimer’s disease: the IWG-2 criteria. Lancet Neurol 2014 ; 13 : 614 -29. OpenUrl CrossRef PubMed Web of Science
• ↵ Walker Z, Moreno E, Thomas A, et al. Clinical usefulness of dopamine transporter SPECT imaging with 123I-FP-CIT in patients with possible dementia with Lewy bodies: randomised study. Br J Psychiatry 2015 ; 206 : 145 -52. OpenUrl Abstract / FREE Full Text
• ↵ Rossor MN, Fox NC, Mummery CJ, et al. The diagnosis of young-onset dementia. Lancet Neurol 2010 ; 9 : 793 -806. OpenUrl CrossRef PubMed Web of Science
• ↵ Ritchie C, Smailagic N, Noel-Storr AH, et al. Plasma and cerebrospinal fluid amyloid beta for the diagnosis of Alzheimer’s disease dementia and other dementias in people with mild cognitive impairment (MCI). Cochrane Database Syst Rev 2014 ; 6 : CD008782 . OpenUrl PubMed
• ↵ Zhang S, Smailagic N, Hyde C, et al. (11)C-PIB-PET for the early diagnosis of Alzheimer’s disease dementia and other dementias in people with mild cognitive impairment (MCI). Cochrane Database Syst Rev 2014 ; 7 : CD010386 . OpenUrl PubMed
• ↵ Robinson L, Gemski A, Abley C, et al. The transition to dementia: individual and family experiences of receiving a diagnosis: a review. Int Psychogeriatr 2011 ; 23 : 1026 -43. OpenUrl CrossRef PubMed
• ↵ National Institute for Health and Care Excellence. Donepezil, galantamine, rivastigmine and mementine for the treatment of Alzheimer’s disease. (Technology appraisal guidance TA217) 2011. http://guidance.nice.org.uk/guidanceTA217 .
• ↵ Birks J. Cholinesterase inhibitors for Alzheimer’s disease. Cochrane Database Syst Rev 2006 ; 1 : CD005593 . OpenUrl PubMed
• ↵ Howard R, McShane R, Lindesay J, et al. Donepezil and memantine for moderate-to-severe Alzheimer’s disease. N Engl J Med 2012 ; 366 : 893 -903. OpenUrl CrossRef PubMed Web of Science
• ↵ Schneider LS, Dagerman KS, Higgins JP, et al. Lack of evidence for the efficacy of memantine in mild Alzheimer disease. Arch Neurol 2011 ; 68 : 991 -8. OpenUrl CrossRef PubMed Web of Science
• ↵ Knapp M, Iemmi V, Romeo R. Dementia care costs and outcomes: a systematic review. Int J Geriatr Psychiatry 2013 ; 28 : 551 -61. OpenUrl CrossRef PubMed
• ↵ Gibson G, Newton L, Pritchard G, et al. The provision of assistive technology products and services for people with dementia in the United Kingdom. Dementia (London) 2014; published online 5 May.
• ↵ Robinson L, Dickinson C, Rousseau N, et al. A systematic review of the effectiveness of advance care planning interventions for people with cognitive impairment and dementia. Age Ageing 2012 ; 41 : 263 -9. OpenUrl Abstract / FREE Full Text
• ↵ Van der Steen JT. Dying with dementia: what we know after more than a decade of research. J Alzheimers Dis 2010 ; 22 : 37 -55. OpenUrl PubMed Web of Science
• ↵ Brodaty H, Green A, Koschera A. Meta-analysis of psychosocial interventions for caregivers of people with dementia. J Am Geriatr Soc 2003 ; 51 : 657 -64. OpenUrl CrossRef PubMed Web of Science
• ↵ Cameron ID, Aggar C, Robinson AL, et al. Assessing and helping carers of older people. BMJ 2011 ; 343 : d5202 . OpenUrl FREE Full Text

Advertisement

Machine Learning for Dementia Prediction: A Systematic Review and Future Research Directions

• Original Paper
• Open access
• Published: 01 February 2023
• Volume 47 , article number  17 , ( 2023 )

Cite this article

You have full access to this open access article

• Ashir Javeed 1 , 2   na1 ,
• Ana Luiza Dallora 2   na1 ,
• Johan Sanmartin Berglund 2 ,
• Arif Ali 3 ,
• Liaqata Ali 4 &
• Peter Anderberg 2 , 5  

15k Accesses

26 Citations

46 Altmetric

Explore all metrics

Nowadays, Artificial Intelligence (AI) and machine learning (ML) have successfully provided automated solutions to numerous real-world problems. Healthcare is one of the most important research areas for ML researchers, with the aim of developing automated disease prediction systems. One of the disease detection problems that AI and ML researchers have focused on is dementia detection using ML methods. Numerous automated diagnostic systems based on ML techniques for early prediction of dementia have been proposed in the literature. Few systematic literature reviews (SLR) have been conducted for dementia prediction based on ML techniques in the past. However, these SLR focused on a single type of data modality for the detection of dementia. Hence, the purpose of this study is to conduct a comprehensive evaluation of ML-based automated diagnostic systems considering different types of data modalities such as images, clinical-features, and voice data. We collected the research articles from 2011 to 2022 using the keywords dementia, machine learning, feature selection, data modalities, and automated diagnostic systems. The selected articles were critically analyzed and discussed. It was observed that image data driven ML models yields promising results in terms of dementia prediction compared to other data modalities, i.e., clinical feature-based data and voice data. Furthermore, this SLR highlighted the limitations of the previously proposed automated methods for dementia and presented future directions to overcome these limitations.

Similar content being viewed by others

Diagnosis of Early Alzheimer’s Disease: Clinical Practice in 2021

Artificial Intelligence for Mental Health and Mental Illnesses: an Overview

Artificial intelligence in disease diagnosis: a systematic literature review, synthesizing framework and future research agenda

Avoid common mistakes on your manuscript.

Introduction

Over a period of time, the advancements made in the field of medical science helped to increase the lifespan in the modern world [ 1 ]. This increased life expectancy raised the prevalence of neurocognitive disorders, affecting a significant part of the older population as well as global economies. In 2010, it was estimated that $604 billion have been spent on dementia patients in the USA alone[ 2 ]. The number of dementia patients is rapidly increasing worldwide, and statistical projections suggest that 135 million people might be affected by dementia by 2050 [ 3 ]. There are several risk factors that contribute to the development of dementia, including aging, head injury, and lifestyle. While age is the most prominent risk factor for dementia; figures suggest that a person at the age of 65 years old has 1–2% risk of developing dementia disease. By the age of 85 years old, this risk can reach to 30% [ 4 ].

Dementia is a mental disorder that is characterized by a progressive deterioration of cognitive functions that can affect daily life activities such as memory, problem solving, visual perception, and the ability to focus on a particular task. Usually, older adults are most vulnerable to dementia, and people take it as an inevitable consequence of aging, which is perhaps the wrong perception. Dementia is not a part of the normal ageing process; however, it should be considered a serious form of cognitive decline that affects your daily life. Actually, the primary cause for the development of dementia is the several diseases and injuries that affect the human brain [ 5 ]. Dementia is ranked on the seventh place in the leading causes of deaths in the world [ 6 ]. Furthermore, it is the major cause of disability and dependency among older people globally [ 6 ]. A change in the person’s ordinary mental functioning and obvious signs of high cognitive deterioration are required for a diagnosis of dementia [ 7 ]. Figure 1 presents the progression of dementia with age.

Progression of dementia disease with ageing

Types of dementia

Dementia is not a single disease, but, it is used as a generic term for several different cognitive disorders. Figure 2 provides the overview of different types of dementia along with the percentage of particular dementia type occurrence in the patients [ 8 ]. To have a better idea about dementia, we have studied common types of dementia for better problem awareness.

Types of dementia disease

Alzheimer’s disease

Alzheimer’s disease (AD) is thought to develop when abnormal amounts of amyloid beta (A \(\beta\) ) build up in the brain, either extracellularly as amyloid plaques, tau proteins or intracellularly as neurofibrillary tangles, affecting neuronal function, connectivity and leading to progressive brain function loss [ 9 ]. This diminished ability to eliminate proteins with ageing is regulated by brain cholesterol [ 10 ] and is linked to other neurodegenerative illnesses [ 11 ]. Except for 1–2% of cases where deterministic genetic anomalies have been discovered, the aetiology of the majority of Alzheimer’s patients remains unexplained [ 12 ]. The amyloid beta (A \(\beta\) ) hypothesis and the cholinergic hypothesis are two competing theories presented to explain the underlying cause of AD [ 13 ].

Vascular dementia

Vascular dementia (VaD) is a subtype of dementia caused by problems with the brain’s blood flow, generally in the form of a series of minor strokes, which results in a slow decline of cognitive capacity [ 14 ]. The VaD refers to a disorder characterized by a complicated mix of cerebrovascular illnesses that result in structural changes in the brain, as a result of strokes and lesions, which lead to cognitive impairment. A chronological relationship between stroke and cognitive impairments is necessary to make the diagnosis [ 15 ]. Ischemic or hemorrhagic infarctions in several brain areas, such as the anterior cerebral artery region, the parietal lobes, or the cingulate gyrus, are associated with VaD. In rare cases, infarcts in the hippocampus or thalamus might cause dementia [ 16 ]. A stroke increases the risk of dementia by 70%, whereas a recent stroke increases the risk by almost 120% [ 17 ]. Brain vascular lesions can also be caused by diffuse cerebrovascular disease, such as small vessel disease [ 18 ]. Risk factors for VaD include age, hypertension, smoking, hypercholesterolemia, diabetes mellitus, cardiovascular disease, and cerebrovascular sickness; geographic origin, genetic proclivity, and past strokes are also risk factors [ 19 ]. Cerebral amyloid angiopathy, which develops when beta amyloid accumulates in the brain, can occasionally lead to vascular dementia.

Lewy body dementia

Lewy body dementia (LBD) is a subtype of dementia characterized by abnormal deposits of the protein alpha-synuclein in the brain. These deposits, known as Lewy bodies, affect brain chemistry, causing problems with thinking, movement, behavior, and mood. Lewy body dementia is one of the most common causes of dementia [ 20 ]. Progressive loss of mental functions, visual hallucinations, as well as changes in alertness and concentration are prevalent in persons with LBD. Other adverse effects include tight muscles, delayed movement, difficulty walking, and tremors, all of which are also signs and symptoms of Parkinson’s disease [ 21 ]. LBD might be difficult to identify. Early LBD symptoms are commonly confused with those of other brain diseases or mental problems. Lewy body dementia can occur alone or in conjunction with other brain disorders [ 22 ]. It is a progressive disorder, which means that symptoms emerge gradually and worsen with time. A timespan of five to eight years is averaged, although it can last anywhere from two to twenty years for certain people [ 23 ]. The rate at which symptoms arise varies greatly from person to person, depending on overall health, age, and the severity of symptoms.

Frontotemporal dementia

Frontotemporal Dementia (FTD) is a subtype of dementia characterized by nerve cell loss in the frontal and temporal lobes of the brain [ 24 ]. As a result, the lobes contract. FTD can have an impact on behavior, attitude, language, and movement. This is one of the most common dementias in people under the age of 65. FTD most commonly affects persons between the ages of 40 and 65; however, it may also afflict young adults and older individuals [ 25 ]. The lobes decrease, and behavior, attitude, language, and mobility can all be affected by FTD. FTD affects both men and women equally. Dissociation from family, extreme oniomania, obscene speech, screaming, and the inability to regulate emotions, behavior, personality, and temperament are examples of social display patterns caused by FTD [ 26 ]. The symptoms of FTD appeared several years prior to visiting a neurologist [ 27 ].

Mixed Dementia (MD)

Mixed dementia occurs, when more than one kind of dementia coexists in a patient, and it is estimated to happen in around 10% of all dementia cases [ 6 ]. AD and VaD dementia are the two subtypes that are most common in MD [ 28 ]. This case is usually associated with factors such as old age, high blood pressure, and brain blood vessel damage [ 29 ]. Because one dementia subtype often predominates, MD is difficult to identify. As a result, the individuals affected by MD are rarely treated and miss out on potentially life-changing medicines. MD can cause symptoms to begin earlier than the actual diagnosis of the disease and spread swiftly to affect the most areas of the brain [ 30 ].

Recently, numerous automated methods have been developed based on machine learning for early the prediction of different diseases [ 31 , 32 , 33 , 34 , 35 , 36 , 37 , 38 , 39 , 40 , 41 , 42 , 43 , 44 , 45 , 46 , 47 , 48 ]. This systematic literature review (SLR) presented hereby, investigates machine learning-based automated diagnostic systems that are designed and developed by scientists to predict dementia and its subtypes, such as AD, VaD, LBD, FTD and MD. We used the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) criteria to conduct this SLR [ 49 , 50 ]. A comprehensive search was conducted to retrieve the research articles that contain ML approaches to predict the development of dementia and its subtypes using three different types of data modalities (images, clinical-variables, voice).

Aim of the study

SLRs are done to synthesize current evidence, to identify gaps in the literature, and to provide the groundwork for future studies [ 51 ]. Previous, SLRs studies have been done on automated diagnostic systems for dementia prediction based on ML approaches, which focused on a single sort of data modality. These SLR investigations did not emphasize the limits of previously published automated approaches for dementia prediction. The SLR presented herein assesses the previously proposed automated diagnostic systems based on deep learning (DL) and ML algorithms for the prediction of dementia and its common subtypes (e.g. AD, VaD, FTD, MD). The aim of this SLR is to analyse and evaluate the performance of automated diagnostic systems for dementia prediction using different data modalities. The main question is decomposed in the following sub-research questions:

What types of ML and DL techniques have been used by researchers to diagnose dementia?

Examine the methods of feature extraction or selection used by the researchers.

Analyze the different performance evaluation measures that are adopted by the researcher to validate the effectiveness of the proposed diagnostic system for demetnia.

Analyze the performance of ML models on various data types.

Identification of weaknesses in previously proposed ML models for dementia prediction.

Flow diagram of PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses)

Article selection

For this SLR study, the research articles were selected based on keywords such as ML, DL, dementia and its subtypes (AD, VaD, FTD, and MD). For the collection of research articles, we conducted an electronic search from different online databases such as ScienceDirect, PubMed, IEEE Xplore Digital Library, Springer, Hindawi, and PLOs, which helped to gather 450 research studies on the specific topic. After reviewing the title and abstract in each study, 120 publications were found to be ineligible for processing, while 330 articles were selected for further processing. Following the deduplication of data, 125 full-text publications were retrieved for further processing after the screening phase of the article selection, with 205 of them being eliminated due to not satisfying the article selection criteria of the screening phase. Finally, 50 research articles were eliminated due to not fulfilling the eligibility criteria for article selection. The final set of selected papers consisted of 75 research papers, among these final selected articles, each of the data modalities (image, clinical-variables, voice) contained 25 papers. After rerunning the database searches in May 2022, no further suitable research article was found for the selection. Figure 3 presents the workflow for article selection, which includes the four PRISMA guidelines-recommended steps such as identification, screening, eligibility, and inclusion [ 49 , 50 ]. In recent years, ML scientists have shown a strong interest in designing and developing ML-based automated diagnostic systems for dementia prediction. Therefore, the number of research articles in this research area has been increased and it can be depicted from Fig. 4 where research articles are published years wise with regarding data modality. The publications utilized in this study were selected based on the following criteria:

Studies that present automated diagnostic systems for dementia and its common subtypes (AD,VaD, FTD, MD).

Studies published between 2011 and 2022.

Studies employing ML approaches for dementia diagnosis.

Studies which have utilized several data modalities.

Studies published in the English language.

Selected research articles which are published from 2011 to 2022 regarding data modality

Machine learning for dementia

Over the years, the increasing use and availability of medical equipment has resulted in a massive collection of electronic health records (EHR) that might be utilized to identify dementia using developing technologies such as ML and DL [ 52 ]. These EHRs are one of the most widely available and used clinical datasets. They are a crucial component of contemporary healthcare delivery, providing rapid access to accurate, up-to-date, comprehensive patient information while also assisting with precise diagnosis and coordinated, efficient care [ 53 ]. Laboratory tests, vital signs, drugs, and other therapies, as well as comorbidities, can be used to identify the people at risk of dementia using the EHRs’ data [ 54 ]. In some situations, patients may also be subjected to costly and invasive treatments such as neuroimaging scans i.e., magnetic resonance imaging (MRI) and position emission tomography (PET)) and cerebrospinal fluid (CSF) collection for biomarker testing [ 55 , 56 , 57 ]. These tests’ findings may also be found in the EHR. According to researchers, such longitudinal clinical EHR data can be used to track the advancement of AD dementia over time [ 58 ]. Recently, several automated diagnostic systems for different diseases, such as Parkinson’s disease [ 59 ], hepatitis [ 47 ], carcinoma [ 41 ], and heart failure [ 60 , 61 , 62 ] prediction have been designed by employing ML and DL techniques. Inspired by this fact, the unmet demand for dementia knowledge, along with the availability of relevant huge datasets, has motivated scientists to investigate the utility of artificial intelligence (AI), which is gaining a prominent role in the area of healthcare innovation [ 63 ]. ML, a subset of AI, can model the relationship between input quantities and clinical outcomes, identify hidden patterns in enormous volumes of data, and draw conclusions or make decisions that help with more accurate clinical decision-making [ 51 ]. However, computational hypotheses generated by ML models must still be confirmed by subject matter experts in order to achieve enough precision for clinical decision-making [ 64 ].

In this SLR, we have included studies that have used ML predictive models (supervised and unsupervised) for dementia prediction and excluded studies that have used statistical methods for cohort summarization and hypothesis testing (e.g., odds ratio, chi-square distribution, Kruskal-Wallis test, and Kappa-Cohen test). Furthermore, we have referenced the data modality-based study [ 65 ] for this literature review, where we have categorized the three data modality types such as image, clinical-variable and voice. Thus, we have studied each modality-based automated diagnostic system for dementia prediction that has been proposed in the past using ML and DL.

This section explains the datasets that were used in the selected research papers for experiments and performance evaluation of the proposed automated diagnostic systems designed by the researchers using ML algorithms for dementia and its subtypes. A total of 61 datasets were studied from the selected research articles. These datasets are compiled from a wide range of organizations and hospitals throughout the world. Only a few datasets are openly available to the public, while others are compiled by researchers from various hospitals and healthcare institutes. We have only included datasets that have been used to diagnose AD, VaD, FTD, MD, and LBD using ML and DL techniques. On the basis of data modality, we have categorised the dataset into three types: images, clinical_variables and voice datasets. The datasets differ in terms of the number of variables (features) and samples. As a result, we examined each modality of the dataset one by one.

Image modality based datasets

There are several image datasets based on brain imaging, such as magnetic resonance imaging (MRI), collected by the researchers for the diagnosis of dementia. From the Table 1 , it can be depicted that Open Access Series of Imaging Studies (OASIS) and Alzheimer’s Disease Neuroimaging Initiative (ADNI) datasets are mostly used by the researchers for the experimental purpose. OASIS aims to make neuroimaging datasets available to the scientific community for free. By gathering and openly disseminating this multimodal dataset produced by the Knight ADRC and its related researchers, they had used different samples and variables of the datasets in their research work. ADNI researchers acquire, validate, and use data such as MRI, PET imaging, genetics, cognitive assessments, CSF, and blood biomarkers as disease predictors. The ADNI website contains research information and data from the North American ADNI project, which includes Alzheimer’s disease patients, people with mild cognitive impairment, and older controls. Table 1 provides us with the following information: dataset_id, dataset name, number of samples in the particular dataset, variables in the dataset, and finally, the type of dementia.

Clinical-variables modality based datasets

Throughout the course of time, the growing usage and availability of medical devices have resulted in an overwhelming collection of clinical EHR data. Furthermore, the patient’s medical history consists of medical tests and clinical records that can be used for the prediction of diseases. Thus, the importance of clinical data emerges as a vital tool for proactive management of disease. The dataset based on clinical variables for dementia consists of medical tests that are used by doctors to check the dementia status in patients, such as the Mini Mental Status Exam (MMSE), the Montreal Cognitive Assessment (MoCA), the Telephone Interview for Cognitive Status (TICS), and the Brief Interview for Mental Status (BIMS). Clinical-variables based datasets consist of information about these medical tests along with patient personal information, i.e., age, sex, and marital status. Hereby, Table 2 provides the information regarding clinical-variables modality-based datasets that are used by the researchers for the design and development of automated diagnostic systems for dementia patients based on ML. Table 2 presents the dataset_id, dataset name, number of samples in the particular dataset, variables in the dataset, and finally the type of dementia.

Voice modality based datasets

Speech analysis is a useful technique for clinical linguists in detecting various types of neurodegenerative disorders affecting the language processing areas. Individuals suffering from Parkinson’s disease (PD, deterioration of voice quality, unstable pitch), Alzheimer’s disease (AD, monotonous pitch), and the non-fluent form of Primary Progressive Aphasia (PPA-NF, hesitant, non-fluent speech) may experience difficulties with prosody, fluency, and voice quality. Besides imaging and clinical-variables data, the researchers employed voice recording data to identify dementia using ML and DL algorithms. The data collection process for voice data varies from dataset to dataset, for example, in a few datasets, patients were requested to answer a prepared set of questions (interview) in a specific time interval. In a few datasets, selected neuropsychological tests were carried out, the description of each neuropsychological test was played and was followed by an answering window. Table 3 presents the dataset_id, dataset name, number of samples in the particular dataset, variables in the dataset, and finally the subtype of dementia.

Data sharing challenges

In this digital era, public health decision-making has grown progressively complicated, and the utilization of data has become critical [ 66 ]. Data are employed at the local level to observe public health and target interventions; at the national scale for resource allocation, prioritization, and planning; and at the global scale for disease burden estimates, progress in health and development measurement, and the containment of evolving global health threats [ 67 , 68 ]. Van Panhuis et al. have adequately described the challenges to exchanging health data [ 69 ]. Based on our initial analysis, we built on this taxonomy to identify the hurdles related to data sharing in global public health, and we have highlighted how they may apply to each typology as given below.

Lack of complete data, lost data, restrictive as well as conflicting data formats, a lack of metadata and standards, a lack of interoperability of datasets (e.g., structure or “language”), and a lack of appropriate analytic solutions are examples of technical barriers encountered by health information management systems.

Individuals and organizations face motivational challenges when it comes to sharing data. These impediments include a lack of incentives, opportunity costs, apprehension about criticism, and disagreements over data usage and access.

The potential and present costs of sharing data are both economic hurdles.

Political obstacles are those that are built into the norms of local health governance and often emerge as regulations and guidelines. They can also entail trust and ownership difficulties.

Legal issues that arise as a result of data collection, analysis, and usage include questions regarding who owns or controls the data, transparency, informed permission, security, privacy, copyright, human rights, damage, and stigma.

Ethical constraints include a lack of perceived reciprocity (i.e., the other side will not disclose data) and proportionality (i.e., deciding not to share data based on an assessment of the risks and benefits). An overall concern is that frameworks, rules, and regulations have not kept up with technological changes that are transforming how data is collected, analyzed, shared, and used.

ML based diagnostic models for dementia: Image modality

In recent years, researchers have designed many ML and DL algorithms for the detection of dementia and its subtypes using MRI images of the brain. For example, Dashtipour et al. [ 70 ] proposed a ML based method for the prediction of Alzheimer’s disease. In their proposed model, they used DL techniques to extract the features from brain images, and for classification purposes, they deployed SVM and bidirectional long short-term memory (BiLSTM). Through their proposed model, they had reported the classification accuracy of 91.28%. Moreover, for early detection of the AD, a DL based approach was proposed by Helaly et al. In their proposed work, they employed convolutional neural networks (CNN). The Alzheimer’s disease spectrum is divided into four phases. Furthermore, different binary medical image classifications were used for each two-pair class of Alzheimer’s disease stages. Two approaches were used to categorize medical images and diagnose Alzheimer’s disease. The first technique employs basic CNN architectures based on 2D and 3D convolution to cope with 2D and 3D structural brain images from the ADNI dataset. They had achieved highly promising accuracies for 2D and 3D multi-class AD stage classification of 93.61% and 95.17%, respectively. The VGG19 pre-trained model had been fine-tuned and obtained an accuracy of 97% for multi-class AD stage classification [ 71 ]. Vandenberghe et al. had proposed a method for binary classification of 18F-flutemetramol PET using ML techniques for AD and mild cognitive impairment (MCI). They had tested whether support vector machines (SVM), a supervised ML technique, can duplicate the assignments made by blindfolded visual readers, as well as which image components had the highest diagnostic value according to SVM and how 18F-fluoromethylamol-based SVM classification compares to structural MRI-based SVM classification in the same cases. Their F-flutemetamol based classifier was able to replicate the assignments obtained by visual read with 100% accuracy [ 72 ]. Odusami et al. proposed a novel method for the detection of early-stage dementia from functional brain changes in MRI using a fine-tuned ResNet-18 network. Their research work presents a DL based technique for predicting MCI, early MCI, late MCI, and Alzheimer’s disease (AD). The ADNI fMRI dataset was used for analysis and consisted of 138 participants. On EMCI vs. AD, LMCI vs. AD, and MCI vs. AD, the fine-tuned ResNet18 network obtained classification accuracy of 99.99%, 99.95%, and 99.95%, respectively [ 73 ]. Zheng et al. had presented a ML based framework for differential diagnosis between VaD and AD using structural MRI features. The least absolute shrinkage and selection operator (LASSO) was then used to build a feature set that was fed into SVM for classification. To ensure unbiased evaluation of model performance, a comparative analysis of classification models was conducted using different ML algorithms to discover which one had better performance in the differential diagnosis between VaD and AD. The diagnostic performance of the classification models was evaluated using quantitative parameters derived from the receiver operating characteristic curve (ROC). The experimental finding had shown that the SVM with RBF performed well for the differential diagnosis of VaD and AD, with sensitivity (SEN), specificity (SPE), and accuracy (ACC) values of 82.65%, 87.17%, and 84.35%, respectively (AUC = 86.10–95%, CI = 0.820–0.902) [ 74 ]. Basheer et al. [ 75 ] had presented an innovative technique by making improvements in capsule network design for the best prediction outcomes. The study used the OASIS dataset with dimensions (373 X 15) to categorize the labels as demented or non-demented. To make the model swifter and more accurate, several optimization functions were performed on the variables, as well as the feature selection procedure. The claims were confirmed by demonstrating the correlation accuracy at various iterations and layers with an allowable accuracy of 92.39%. L. K. Leong and A. A. Abdullah had proposed a method for the prediction of AD based on ML techniques with the Boruta algorithm as a feature selection method. According to the Boruta algorithm, Random Forest Grid Search Cross Validation (RF GSCV) outperformed other 12 ML models, including conventional and fine-tuned models, with 94.39% accuracy, 88.24% sensitivity, 100.00% specificity, and 94.44% AUC even for the small OASIS-2 longitudinal MRI dataset [ 76 ]. Battineni et al. had presented a SVM based ML model for the prediction of dementia. Their proposed model had achieved an accuracy and precision of 68.75% and 64.18% using the OASIS-2 dataset [ 77 ]. Mathotaarachchi et al. had analyzed the amyloid imaging using ML approaches for the detection of dementia. To overcome the inherent unfavorable and imbalance proportions between persons with stable and progressing moderate cognitive impairment in a short observation period. The innovative method had achieved 84.00% accuracy and an AUC of 91.00% for the ROC [ 78 ]. Aruna and Chitra had presented a ML approach for the identification of dementia from MRI images, where they had deployed Independent Component Analysis (ICA) to extract the features from the images, and for classification purposes, SVM with different kernels is used. Through their proposed method, they had obtained an accuracy of 90.24% [ 79 ] (Fig. 5 ).

Accuracy comparison of different ML models based on image modality

Supervised ML techniques and CNNs were examined by Herzog and Magoulas. They had achieved the accuracy of 92.5% and 75.0% for NC vs EMCI, 93.0% and 90.5% for NC vs. AD, respectively [ 80 ]. Battineni et al. had comprehensive applied ML model on MRI to predict Alzheimer’s disease (AD) in older subjects, and they had proposed two ML models for AD detection. In the first trial, manual feature selection was utilized for model training, and ANN produced the highest AUC of 81.20% by ROC. The NB had earned the greatest AUC of 94.20% by ROC in the second trial, which included wrapping approaches for the automated feature selection procedure [ 81 ]. Ma et al. had conducted a study where they compared feature-engineered and non-feature-engineered ML methods for blinded clinical evaluation for dementia of Alzheimer’s type classification using FDG-PET. The highest accuracy of 84.20% was obtained through CNN’s [ 82 ]. Bidani et al. had presented a novel approach in the field of DL that combines both the deep convolutional neural network (DCNN) model and the transfer learning model to detect and classify dementia. When the features were retrieved, the dementia detection and classification strategy from brain MRI images using the DCNN model provided an improved classification accuracy of 81.94%. The transfer learning model, on the other hand, had achieved an accuracy of 68.13% [ 83 ].

Moscoso et al. had designed a predictive model for the prediction of Alzheimer’s disease using MRI images. Their proposed model had obtained the highest accuracy of 84.00% [ 84 ]. Khan and Zubair had presented an improved multi-modal based ML approach for the prognosis of AD. Their proposed model had a five-stage ML pipeline, where each stage was further categorized into different sub-levels. Their proposed model had reported the highest accuracy of 86.84% using RF [ 85 ]. Mohammed et al. had evaluated the two CNN models (AlexNet and ResNet-50) and hybrid DL/ML approaches (AlexNet+SVM and ResNet-50+SVM) for AD diagnosis using the OASIS dataset. They had found that RF algorithm had attained an overall accuracy of 94%, as well as precision, recall, and F1 scores of 93%, 98%, and 96%, respectively [ 86 ]. Salvatore et al. had developed a ML method for early AD diagnosis using magnetic resonance imaging indicators. In their proposed ML model, they used PCA for extracting features from the images and SVM for the classification of dementia. They had achieved a classification accuracy of 76% using a 20-fold cross validation scheme [ 87 ]. Katako et al. had identified the AD related FDGPET pattern that is also found in LBD and Parkinson’s disease dementia using ML approaches. They studied different ML algorithms, but SVM with an iterative single data algorithm produced the best performance, i.e., sensitivity 84.00%, specificity 95.00% through 10-fold cross-validation [ 88 ]. Gray et al. had presented a system in which RF proximities were utilized to learn a low-dimensional manifold from labelled training data and then infer the clinical labels of test data that translated to this space. Their proposed model, voxel-based (FDG-PET), obtained an accuracy of 87.9% using ten-fold cross-validation [ 89 ]. Table 4 provides the overall performance evaluation of the ML models that were presented by the researchers for the prediction of dementia and its subtypes by using image data as a modality.

ML based diagnostic models for dementia: Clinical-variable modality

Aside from image-based ML techniques for dementia prediction, several research studies have utilized clinical-variable data with ML algorithms to predict dementia and its subtypes. For instance, Chiu et al. had designed a screening instrument to detect MCI and dementia using ML techniques. They had developed a questionnaire to assist neurologists and neuropsychologists in the screening of MCI and dementia. The contribution of 45 items that matched the patient’s replies to questions was ranked using feature selection through information gain (IG). Among the 45 items, 12 were ranked the highest in feature selection. The ROC analysis showed that AUC in test group was 94.00% [ 96 ]. Stamate et al. had developed a framework for the prediction of MCI and dementia. Their proposed framework was based on the ReliefF approach paired with statistical permutation tests for feature selection, model training, tweaking, and testing using ML algorithms such as RF, SVM, Gaussian Processes, Stochastic Gradient Boosting, and eXtreme Gradient Boosting. The stability of model performances was studied using computationally expensive Monte Carlo simulations, and the results of their proposed framework were given as for dementia detection, the accuracy was 88.00%, sensitivity was 93.00%, and the specificity was 94.00%, whereas moderate cognitive impairment had a sensitivity of 86.00% and a specificity of 90% [ 97 ]. Stamate et al. developed a system for detecting dementia subtypes (AD) in blood utilizing DL and other supervised ML approaches such as RF and extreme gradient boosting. The AUC for the proposed DL method was 85% (0.80–0.89), for XGBoost it was 88% (0.86–0.89), and for RF it was 85% (0.83–0.87). In comparison, CSF measurements of amyloid, p-tau, and t-tau (together with age and gender) gave AUC values of 78%, 83%, and 87%, respectively, by using the XGBoost [ 98 ]. Bansal1 et al. had performed the comparative analysis of the different ML methods for the detection of dementia using clinical-variables. In their experiments, they exploited the performance of four ML models, such as J48, NB, RF, and multilayer perceptrons. From the results of experiments, they had concluded that j48 outperformed the rest of the ML models for the detection of dementia [ 99 ]. Nori et al. had experimented the lasso algorithm on a big dataset of patient and identify the 50 variables by ML model with an AUC of 69.30% [ 100 ]. Alam et al. [ 101 ]used signal processing on wearable sensor data streams (e.g., electrodermal activity (EDA), photoplethysmogram (PPG), and accelerometer (ACC)) and machine learning techniques to measure cognitive deficits and their relationship with functional health deterioration.

Gurevich et al. had used SVM and neuropsychological test for the classification of AD from other causes of cognitive impairment. The highest classification accuracy they had achieved through their proposed method was 89.00% [ 102 ]. Karaglani et al. had proposed a ML based automated diagnosis system for AD by using blood-based biosignatures. In their proposed method, they used mRNA-based statistically equivalent signatures for feature ranking and a RF model for classification. Their proposed automated diagnosis system had reported the accuracy of 84.60% using RF [ 103 ]. Ryzhikova et al. had analyzed cerebrospinal fluid for the diagnosis of AD by using ML algorithms. For classification purposes, artificial neural networks (ANN) and SVM discriminant analysis (SVM-DA) statistical methods were applied, with the best findings allowing for the distinguishing of AD and HC participants with 84.00% sensitivity and specificity. The proposed classification models have a high discriminative power, implying that the technique has a lot of potential for AD diagnosis [ 104 ]. Cho and Chen had designed a double layer dementia diagnosis system based on ML where fuzzy cognitive maps (FCMs) and probability neural networks (PNNs) were used to provide initial diagnoses at the base layer, and Bayesian networks (BNs) were used to provide final diagnoses at the top layer. Diagnosis results, “proposed treatment,” and “no treatment required” might be used to provide medical institutions with self-testing or secondary dementia diagnosis. The highest accuracy reported by their proposed system was 83.00% [ 105 ]. Facal et al. had studied the role of cognitive reserve in the conversion from MCI to dementia using ML. Nine ML classification algorithms were tried in their study, and seven relevant performance parameters were generated to assess the prediction accuracy for converted and non-converted individuals. The use of ML algorithms on socio-demographic, basic health, and CR proxy data allowed for the prediction of dementia conversion. The Gradient Boosting Classifier (ACC = 0.93; F1 = 0.86 and Cohen’s kappa = 0.82) and RF Classifier (ACC = 92%; F1 = 0.79 and Cohen’s kappa = 0.71) performed the best [ 106 ]. Jin et al. had proposed automatic classification of dementia from learning of clinical consensus diagnosis in India using ML techniques. All viable ML models exhibited remarkable discriminative skills (AUC >90%) as well as comparable accuracy and specificity (both around 95%). The SVM model beat other ML models by obtaining the highest sensitivity (0.81), F1 score (0.72), kappa (.70, showing strong agreement), and accuracy (second highest) (0.65). As a consequence, the SVM was chosen as the best model in their research work [ 107 ]. James et al. had evaluated the performance of ML algorithms for predicting the progression of dementia in memory clinic patients. According to their findings, ML algorithms outperformed humans in predicting incident all-cause dementia within two years. Using all 258 variables, the gradient-boosted trees approach had an overall accuracy of 92% , sensitivity of 0.45, specificity of 0.97, and an AUC of 0.92. Analysis of variable significance had indicated that just 6 variables were necessary for ML algorithms to attain an accuracy of 91% and an AUC of at least 89.00% [ 108 ]. Bougea et al. had investigated the effectiveness of logistic regression (LR), K-nearest neighbours (K-NNs), SVM, the Naive Bayes classifier, and the Ensemble Model to correctly predict PDD or DLB. The K-NN classification model exhibited an overall accuracy of 91.2% based on 15 top clinical and cognitive scores, with 96.42% sensitivity and 81% specificity in distinguishing between DLB and PDD. Based on the 15 best characteristics, the binomial logistic regression classification model had attained an accuracy of 87.5%, with 93.93% sensitivity and 87% specificity. Based on the 15 best characteristics, the SVM classification model had achieved an accuracy of 84.6% of overall instances, 90.62% sensitivity, and 78.58% specificity. A model based on NB classification obtained an accuracy of 82.05%, sensitivity of 93.10%, and a specificity of 74.41%. Finally, an ensemble model, which was constructed by combining the separate ones, attained 89.74% accuracy, 93.75% sensitivity, and 85.73% specificity [ 109 ] (Fig. 6 ).

Accuracy comparison of different ML models based on clinical-variable modality

Salem et al. had presented a regression-based ML model for the prediction of dementia. In their proposed method, they had investigated ML approaches for unbalanced learning. In their suggested supervised ML approach, they started by intentionally oversampling the minority class and undersampling the majority class, in order to reduce the bias of the ML model to be trained on the dataset. Furthermore, they had deployed cost-sensitive strategies to penalize the ML models when an instance was misclassified in the minority class. According to their findings, the balanced RF was the most resilient probabilistic model (with just 20 features/variables) with an F1 score of 0.82, a G-Mean of 0.88, and an AUC of 0.88 using ROC. With a F1-score of 0.74 and an AUC of 0.80 by ROC, the calibrated-weighted SVM was their top classification model for the same number of features [ 110 ]. Gutierrez et al. had designed an automated diagnosis system for the detection of AD and FTD by using feature engineering and genetic algorithms. Their proposed system had obtained the accuracy of 84% [ 111 ]. Mirzaei and Adeli had analyzed the state-of-the-art ML techniques used for the detection and classification of AD [ 112 ]. Hsiu et al. had studied ML algorithms for early identification of cognitive impairment. Their proposed model had obtained the accuracy of 70.32% by threefold cross-validation scheme [ 113 ]. Several classification models were constructed using various ML and feature selection methodologies to automate MCI detection using gait biomarkers. They had demonstrated, however, that dual-task walking differentiated between MCI and CN individuals. The ML model used for MCI pre-screening based on inertial sensor-derived gait biomarkers achieved 71.67% accuracy and 83.33% sensitivity, respectively, as reported by Shahzad et al. [ 114 ]. Hane et al. investigated the use of deidentified clinical notes acquired from multiple hospital systems over a 10-year period to enhance retrospective ML models predicting the risk of developing AD. The AUC improved from 85.00% to 94.00% by utilizing clinical notes, and the positive predictive value (PPV) rose from 45.07% (25,245/56,018) to 68.32% (14,153/20,717) in the model at the beginning of disease [ 115 ]. Table 5 provides the overall performance evaluation of the ML models that were presented by the researchers for the prediction of dementia and its subtypes by using clinical-variable data as a modality.

ML based diagnostic models for dementia: Voice modality

Similar to the image and clinical-variable modalities, researchers had also developed automated diagnostic systems based on voice data for the prediction of dementia. Hereby, we have reviewed the research work done by the scientists in detail. For example, Chlasta and Wolk had worked on the computer-based automated screening of dementia patients by spontaneous speech analysis using DL and ML techniques. In their work, they used neural networks to extract the features from the voice data; the extracted features were then fed into a linear SVM for classification purposes. Their SVM model had obtained the accuracy of 59.1% while CNN based ML model had reported the accuracy of 63.6% [ 121 ]. Chien et al. had presented an ML model for the assessment of AD using speech data. Their suggested model included a feature sequence that was used to extract the features from the raw audio data, as well as a recurrent neural network (RNN) for classification. Their proposed ML model had reported an accuracy of 83.80% based on the ROC curve [ 122 ]. Shimoda et al. had designed an ML model that identified the risk of dementia based on the voice feature in telephone conversations. Extreme gradient boosting (XGBoost), RF, and LR based ML models were used, with each audio file serving as one observation. The predictive performance of the constructed ML models was tested by characterizing the ROC curve and determining the AUC, sensitivity, and specificity [ 123 ]. Nishikawa et al. had developed an ensemble discriminating system based on a classifier with statistical acoustic characteristics and a neural network of transformer models, with an F1-score of 90.70% [ 124 ]. Liu et al. had introduced a new technique for recognizing Alzheimer’s disease that used spectrogram features derived from speech data, which aided families in comprehending the illness development of patients at an earlier stage, allowing them to take preventive measures. They used ML techniques to diagnose AD using speech data collected from older adults who displayed the attributes described in the speech. Their proposed method had obtained the maximum accuracy of 84.40% based on LogisticRegressionCV [ 125 ]. Searle et al. had created a ML model to assess spontaneous speech, which might potentially give an efficient diagnostic tool for earlier AD detection. Their suggested model was a fundamental Term Frequency-Inverse Document Frequency (TF-IDF) vectorizer as input into an SVM model, and the top performing models were a pre-trained transformer-based model ’DistilBERT’ when used as an embedding layer into simple linear models. The proposed model had obtained the highest accuracy of 82.00% [ 126 ]. Zhu et al. had suggested an ML model that employed the speech pause as an effective biomarker in dementia detection, with the purpose of reducing the detection, model’s confidence levels by adding perturbation to the speech pauses of the testing samples. They next investigated the impact of the perturbation in training data on the detection model using an adversarial training technique. The proposed model had achieved an accuracy of 84.00% [ 127 ]. Ossewaarde et al. had proposed ML model based on SVM for the classification of spontaneous speech of individuals with dementia based on automatic prosody analysis. Their findings suggest that the classifier can distinguish some dementia types (PPA-NF, AD), but not others (PD) [ 128 ]. Xue et al. had developed an ML model based on DL for the detection of dementia by using voice recordings. In their ML model, long short-term memory (LSTM) network and the convolutional neural network (CNN) utilized audio recordings to categorize whether the recording contained a participant with either NC or only DE and to discriminate between recordings belonging to those with DE and those without DE (i.e., NDE (NC+MCI)) [ 129 ]. Weiner et al. had presented two pipelines of feature extraction for dementia detection: the manual pipeline used manual transcriptions, while the fully automatic pipeline used transcriptions created by automatic speech recognition (ASR). The acoustic and linguistic features that they had extracted need no language specific tools other than the ASR system. Using these two different feature extraction pipelines, they had automatically detect dementia [ 130 ] (Fig. 7 ).

Accuracy comparison of different ML models based on voice modality

Furthermore, Sadeghian et al. had presented the empirical evidence that a combination of acoustic features from speech, linguistic features were extracted from an automatically determined transcription of the speech including punctuation, and results of a mini mental state exam (MMSE) had achieved strong discrimination between subjects with a probable AD versus matched normal controls [ 131 ]. Khodabakhsh et al. had evaluated the linguistic and prosodic characteristics in Turkish conversational language for the identification of AD. Their research suggested that prosodic characteristics outperformed linguistic features by a wide margin. Three of the prosodic features had helped to achieve a classification accuracy of more than 80%, However, their feature fusion experiments did not improve classification performance any more [ 132 ]. Edwards et al. had analyzed the text data at both the word level and phoneme level, which leads to the best-performing system in combination with audio features. Thus, the proposed system was both multi-modal (audio and text) and multi-scale (word and phoneme levels). Experiments with larger neural language models had not resulted in improvement, given the small amount of text data available [ 133 ]. Kumar et al. had identified speech features relevant in predicting AD based on ML. They had deployed neural network for the classification and obtained the accuracy of 92.05% [ 134 ]. Ossewaarde et al. had built ML model based on SVM for the classification from spontaneous speech of individuals with dementia by using automatic prosody [ 128 ]. Luz et al. had developed an ML approach for analyzing patient speech in dialogue for dementia identification. They had designed a prediction model, and the suggested strategy leveraged additive logistic regression (ML boosting method) on content-free data gathered through dialogical interaction. Their proposed model obtained the accuracy of 86.50% [ 135 ]. Sysed et al. had designed a multimodal system that identified linguistic and paralinguistic traits of dementia using an automated screening tool. Their proposed system had used bag-of-deep-feature for feature selection and ensemble model for classification [ 136 ]. Moreover, Sarawgi et al. had used multimodal inductive transfer learning for AD detection and severity. Their proposed system further achieved state-of-the-art AD classification accuracy of 88.0% when evaluated on the full benchmark DementiaBank Pitt database. Table 6 provides the overall performance evaluation of the ML models that were presented by the researchers for the prediction of dementia and its subtypes by using voice-modality data.

In this SLR, we examined the research work that employed ML and DL algorithms to analyze clinical data in order to identify variables that might help predict dementia. We studied 75 research articles that were published in the last 10 years that used image, clinical-variable, and voice data to predict dementia and its subtypes. Nowadays, the healthcare industry creates a vast quantity of data on patients’ health; this data is used by researchers to enhance individual health by utilizing developing technologies such as ML and DL. As a result, researchers can not only distinguish dementia patients from healthy people with high accuracy, but also forecast the disease progression of MCI patients.

Accuracy comparison of ML models based on data modality

Therefore, researchers have expressed a strong interest in designing and developing automated diagnostic systems based on ML and DL techniques. As seen in Fig. 4 ., there has been an exponential increase in the number of such research publications that use ML algorithms for dementia prediction and detection in the previous four years. We investigated the selected papers using significant performance assessment criteria for ML and DL approaches such as data attributes, computational methodologies, and study emphasis. In this SLR, we have uncovered research gaps in the present literature as well as anticipated future research opportunities. Additionally, in Fig. 8 model comparison, we examined the performance of multiple ML algorithms for dementia prediction based on three types of data modalities: image, clinical-variable, and voice. The accuracy gained by image-based ML algorithms is higher when compared to clinical-variable and voice modalities, as shown in Fig. 8 model comparison. Moreover, the researchers’ suggested SVM, RF, and ANN-based ML techniques outperformed the rest of the ML algorithms in terms of performance. According to Fig. 8 model comparison, voice modality-based ML models show worse accuracy when compared to image and clinical-variable modality data. As a result, there is still a performance gap for researchers to close in order to improve the performance of ML algorithms for the prediction of dementia using voice data. Hence, researchers have shown a strong interest in the creation of automated diagnosis systems for dementia prediction utilizing speech data and ML algorithms, as illustrated in Fig. 4 .

Sensitivity and specificity comparison of ML based on modality

The ML and DL models are likely prone to problems such as poor quality of data, poor selection of ML model, Bias Variance tradeoff and training too complex models. Thus, scientists have developed various evaluation metrics (i.e., ROC, AUC, MCC, F1-score, K-fold) and methods to avoid these problems. The data is a crucial element in ML because ML models work only with numeric data; therefore, poor data quality results in lower performance of ML models. Moreover, imbalance classes in the dataset also cause the bias results from the ML models. Thus, this problem can be overcome by oversampling or undersampling the training data. There are different techniques that are used by the AI engineers for oversampling, such as random oversampling and the synthetic minority oversampling technique (SMOTE). To evaluate the bias researchers’ work, use sensitivity and specificity as an evaluation metric to measure the bias of the ML model. Higher values of sensitivity and specificity means model is free from the biasness while having either one parameter value higher and other one is lower means there is biasness exist. Thus, we have also studied the sensitivity and specificity, along with the accuracy, of the previously proposed ML models for dementia prediction. Figure 9 Comparison provides a brief description of the sensitivity and specificity of the ML models for the detection of dementia based on different data modalities. From Fig. 9 , we can observe that ML models have higher values for sensitivity and specificity when using image data as compared to clinical-variable and voice modality data. In comparison to accuracy from Fig. 8 to sensitivity and specificity from Fig. 9 , we have noted that the results obtained from image based modality are more reliable and precise using ML and DL algorithms in spite of clinical and voice modality.

Furthermore, the correlation between sensitivity and specificity would help us understand the efficacy of the ML models, which are designed for automated disease prediction. The mathematical terms “sensitivity” and “specificity” indicate the accuracy of a test that reports the presence or absence of a disease. Individuals who meet the requirement are labelled “positive,” while those who do not are considered “negative”. The chance of a positive test, conditioned on being actually positive, is referred to as sensitivity (the true positive rate), while specificity (true negative rate) is the likelihood of a negative test if it is actually negative. Sensitivity and specificity are inversely proportional, which means that as sensitivity rises, specificity falls, and vice versa. Mathematically, sensitivity and specificity are given as:

On the other hand, accuracy is a ratio of number of correct assessments / number of all assessments. The proportion of genuine positive outcomes (both true positive and true negative) in the selected population is represented by the numerical value of accuracy. The test result is accurate 99% of the time, whether positive or negative. For the most part, this is right. However, it is worth noting that the equation of accuracy means that even if both sensitivity and specificity are high, say 99%, this does not imply that the test’s accuracy is also high. In addition to sensitivity and specificity, accuracy is determined by the prevalence of the illness in the target population. A diagnosis for a rare ailment in the target group may have high sensitivity and specificity but low accuracy. However, for a balanced dataset, ML models with higher sensitivity and specificity result in higher accuracy. Hence, accuracy must be interpreted carefully. The mathematical formula for accuracy is given as:

where, TP stands for the number of true positives, FP stands for the number of false positives, TN stands for the true negative, and FN stands for the false negative.

Accuracy comparison of ML models along with number of sample in the dataset based on data modality

We classified all datasets that were used by researchers to test the performance of their proposed ML models for the prediction of dementia (AD, VaD, MCI, and FTD) into three types: image, clinical-variable, and voice. A total of 61 datasets were examined in terms of the number of samples and variables in the datasets. In image modality datasets from the Table 1 , it can be observed that the ADNI dataset has a significant number of samples, which is 750, while the NINDS-AIREN dataset has more variables as compared to the rest of the datasets in the image modality data. Moreover, from the Table 2 of clinical-variable modality datasets, it can be noticed that the ADRD dataset has the highest number of samples (44945) as compared to the rest of the dataset, while the Raman spectral dataset has the highest number of variables (366). In the last, Table 3 of voice data modality elaborated the dataset of voice modality where FHS dataset has highest number of samples of 5449 while VBSD dataset had highest variables of 254 as compared to rest of the datasets in voice modality. The type of data and the size of the dataset are two important factors that have a significant influence on the performance of ML models. Thus, we have also studied this factor by comparing the accuracy along with the number of samples in the dataset with respect to data modalities. From Fig. 10 , it can be observed that the majority of the ML models that used image data have higher accuracy along with a higher number of samples in the dataset. There are few ML models that show poor performance when the number of samples in the dataset is large. While, clinical-variable and voice modalities show prominent performance when the number of samples in the dataset is small.

Overall percentage of ML models used in the selected research articles regardless of data modality

Moreover, we examined the effectiveness of ML classifiers utilized by the researchers in their proposed automated diagnostic systems for dementia prediction and classification. According to the selected studies of this SLR, SVM is the most commonly used ML classifier by researchers for the classification of patients and normal subjects using three data modalities (i.e., image, clinical-variable, voice), RF is the second most commonly used ML classifier by researchers, and CNN is the third most commonly used ML classifier by researchers. It can be observed from the Fig. 11 . SVMs are the most powerful tools for the binary classification task, along with RF. From Fig. 8 , we can see that SVM also obtained the highest average accuracy based on three types of data modalities. Hence, this factor also encourages the scientists to employ SVM as a binary classifier for dementia prediction or other disease prediction systems. From Fig. 11 , we can observe the percentage of other ML classifiers that were used by the researchers in selected research articles for the automated diagnosis of dementia.

There are several evaluation metrics that are used for the performance assessment of ML models, such as F1score, AUC, ROC, Matthew’s correlation coefficient (MCC), cross-validation, K-fold, specificity, sensitivity, and accuracy. Each evaluation metric has its own pros and cons. Thus, the selection of appropriate evaluation metrics for the assessment of the ML model is essential to understanding its efficiency and performance. For instance, when data plays a vital role in ML models for decision-making and a dataset has unbalanced classes, it may be possible that results from the ML predictive model might be biassed due to the unbalanced nature of the data in the dataset. Thus, here evaluation metrics help to eliminate the factor of biasness in the results, i.e., the k-fold. The F1-score evolution metric is suitable for the classification of multiple classes in the dataset. while ROC tells us how well the ML model can differentiate binary classes. As a result, AUC and ROC reveal how effectively the probabilities from the positive classes are separated from the probabilities from the negative classes. From Fig. 12 , it can be depicted that cross validation is mostly used in the studies that were selected for this SLR to evaluate the performance of proposed ML models. MCC is the second most used evaluation metric, while ROC is in third place. The proposition of other evaluation metrics used by the researchers to validate the efficiency of their proposed ML models can be observed from Fig. 12 .

Overall percentage of evaluation metrics of ML models used by the researchers in the selected research articles

Limitations in the previously proposed ML models

ML algorithms have been effectively applied to a broad range of real-world challenges, including banking, cybersecurity, transportation, and robots. They do, however, have fundamental limitations that make them inappropriate for every problem. In the clinical domain, researchers have concentrated on the supervised learning approach, developing various automated diagnostics for AD, MCI, and dementia prediction using supervised machine algorithms. From the Figs.  8 and 11 , It can be noticed that supervised ML classifiers are mostly used by the researchers in the selected past research articles. Because supervised machine learning approaches have various limitations, automated diagnostic methods for dementia prediction based on supervised techniques suffer from some, if not all, of these constraints. In this part, we have examined the drawbacks of supervised ML-based techniques for dementia prediction, which are as follows:

The model overfitting problem affects the performance of ML models. As previously indicated, several researchers have used the k-fold cross-validation approach to evaluate the efficacy of their constructed diagnostic system. However, because of data leaks, it may result in highly biassed findings.

To deal with problem of imbalance classes in the dataset, Researchers and scientists had devised several techniques to eliminate the problem of imbalance classes such as random oversampling example (ROSE), synthetic minority over-sampling technique (SMOTE) and random over sampling (ROS) etc. Unfortunately, in the selected study, the researchers had not considered this factor to deal with the problem of imbalanced classes in the dataset that cause problems of bias.

Supervised ML models require training on a dataset; nevertheless, training on a large quantity of significant data is a hard and time-consuming job, especially for slow learning algorithms like kNN.

For training and testing of the ML models, researchers had used different data partitioning methods, which resulted in inconsistent comparisons of accuracy and other evaluation metrics among the proposed ML models for dementia prediction. Thus, standard data partition schemes should be adopted (holdout) for the comparison of ML models developed by the researcher for dementia prediction.

Another challenge with ML-based automated diagnostic systems for dementia is the time complexity of the proposed ML algorithms. The time complexity means the overall time require to complete all the computational tasks by the ML model for making a prediction. The ML model can forecast results only after it has been trained on the training data, which takes time to analyze. Furthermore, ML models include a large number of parameters that must be manually modified in the case of supervised learning. As a result, it takes a significant amount of effort and time to fine-tune the hyperparameters of the ML model in order to get higher performance.

DL technology has demonstrated cutting-edge performances for the prediction of various diseases in the recent years. However, DL technology needs a massive quantity of data for model training, which is a time-consuming and tough task. Due to the complexity of data models, training is quite costly. Furthermore, DL necessitates the use of pricey GPUs and hundreds of workstations, which are not effective in terms of economics.

Future research directions

In recent years, several ML models have been presented for the prediction of AD and MCI; nevertheless, there are still certain areas that need to be explored by academics and experts. In this section, we have discussed different research areas and the future prospects of ML algorithms for dementia detection. We infer from this study that the following major parameters have a role in the efficient identification of dementia and its forms.

Data is extremely important in the case of ML-based automated detection of dementia, especially when DL models are considered. Many of the publicly available datasets, however, are modest in size. But future research should concentrate on gathering a huge number of samples for the datasets. In this SLR, we studied ML-based automated diagnostic systems for dementia prediction using three different kinds of data modalities (image, clinical_variable, voice). From Fig. 10 , it can be observed that only the image modality based ML model obtained the higher accuracy along with the large size of the dataset, while the voice modality based ML model obtained the higher accuracy on a small dataset. Thus, for the researchers, there is still room available for designing and developing the automated prediction of dementia and its sub-types by using voice data. Therefore, the interest of researchers have been tremendously raised for the development of automated diagnostic systems for dementia prediction using voice data modality and this trend can be confirmed from the Fig. 4 . There is still a lot room available for the improvement in design and construction of automated diagnostic systems for the dementia using clinical-variable data modality for the researchers. Because, the ML model was developed in the past using clinical-variable data, it displays mix performance by using clinical_variable modality, i.e., when the number of samples is lower in the dataset, the ML shows lower accuracy. Thus. In the future, we need to increase the number of samples in the dataset so that we have larger datasets for experimental purposes and the designed ML model can be effectively evaluated.

In selected studies of this SLR, the majority of ML algorithms belong to the supervised category of learning. While few researchers used an unsupervised ML approach for the prediction of dementia and its subtypes, Altough, unspervsied learning approaches suffer from the limitation such as less accuracy, more expensive in term of computational etc. Therefore, it will encourage scientists and researchers to design and construct new techniques and methods using supervised ML algorithms that are more precise and accurate for the prediction of dementia and its subtypes. Moreover, in this SLR, we have analysed the various ML models based on three data modalities (image, clinical-variable, and voice), and we have comprehensively compared previously proposed ML-based systems in terms of various evaluation metrics, but with different data modalities, it would be suggested that multimodal processing techniques based on ML would provide more reliable and efficient results. Hence, in the future, researchers should exploit multimodal approaches based on ML for a better prediction of dementia and its subtypes.

In contrast to earlier SLR studies that examined numerous ML techniques proposed for the automated diagnosis of dementia and its subtypes (AD, VaD, FTD, and MCI) using one type of data modality, this study reviewed ML methods for dementia considering different types of data modalities such as image data, clinical variables, and voice data. The research articles published from 2011 to 2022 were gathered using different databases. It was pointed out that ML approaches based on image data modality has shown better performance compared with ML methods trained on clinical variables based data and voice data modality. Furthermore, this study critically evaluated the previously proposed methods and highlighted limitations in these methods. To overcome these limitations, this study presented future research directions in the domain of automated dementia prediction using ML approaches. We hope that this SLR will be helpful for AI and ML researchers and medical practitioners who are working in the domain of automated diagnostic systems for dementia prediction.

Data Availability

Not applicable.

Code Availability

Menéndez, G.: La revolución de la longevidad: cambio tecnológico, envejecimiento poblacional y transformación cultural. Revista de Ciencias Sociales 30 (41), 159–178 (2017)

Prince, M.J., Wimo, A., Guerchet, M.M., Ali, G.C., Wu, Y.-T., Prina, M.: World alzheimer report 2015-the global impact of dementia: An analysis of prevalence, incidence, cost and trends (2015)

Vrijsen, J., Matulessij, T., Joxhorst, T., de Rooij, S.E., Smidt, N.: Knowledge, health beliefs and attitudes towards dementia and dementia risk reduction among the dutch general population: a cross-sectional study. BMC public health 21 (1), 1–11 (2021)

Widiger, T.A., Costa, P.T., Association, A.P., et al : Personality Disorders and the Five-factor Model of Personality. JSTOR, (2013)

Lo, R.Y.: The borderland between normal aging and dementia. Tzu-Chi Medical Journal 29 (2), 65 (2017)

WHO: Dementia. World Health Organization. https://www.who.int/news-room/fact-sheets/detail/dementia

Budson, A.E., Solomon, P.R.: Memory Loss E-book: A Practical Guide for Clinicians. Elsevier Health Sciences, (2011)

friendly wyoming, D.: Types of dementia. Dementia friendly wyoming. https://www.dfwsheridan.org/types-dementia

Tackenberg, C., Kulic, L., Nitsch, R.M.: Familial alzheimer’s disease mutations at position 22 of the amyloid \(\beta\) -peptide sequence differentially affect synaptic loss, tau phosphorylation and neuronal cell death in an ex vivo system. PloS one 15 (9), 0239584 (2020)

Wang, H., Kulas, J.A., Wang, C., Holtzman, D.M., Ferris, H.A., Hansen, S.B.: Regulation of beta-amyloid production in neurons by astrocyte-derived cholesterol. Proceedings of the National Academy of Sciences 118 (33) (2021)

Vilchez, D., Saez, I., Dillin, A.: The role of protein clearance mechanisms in organismal ageing and age-related diseases. Nature communications 5 (1), 1–13 (2014)

Breijyeh, Z., Karaman, R.: Comprehensive review on alzheimer’s disease: Causes and treatment. Molecules 25 (24), 5789 (2020)

Long, J.M., Holtzman, D.M.: Alzheimer disease: an update on pathobiology and treatment strategies. Cell 179 (2), 312–339 (2019)

Román, G.C.: Vascular dementia may be the most common form of dementia in the elderly. Journal of the neurological sciences 203 , 7–10 (2002)

Gold, G., Bouras, C., Canuto, A., Bergallo, M.F., Herrmann, F.R., Hof, P.R., Mayor, P.-A., Michel, J.-P., Giannakopoulos, P.: Clinicopathological validation study of four sets of clinical criteria for vascular dementia. American Journal of Psychiatry 159 (1), 82–87 (2002)

Lina, R.: Atrial fibrillation in aging; methodological aspects and the relation to dementia and cerebral vascular disease (2022)

Kuźma, E., Lourida, I., Moore, S.F., Levine, D.A., Ukoumunne, O.C., Llewellyn, D.J.: Stroke and dementia risk: a systematic review and meta-analysis. Alzheimer’s & Dementia 14 (11), 1416–1426 (2018)

Tay, J., Morris, R.G., Tuladhar, A.M., Husain, M., de Leeuw, F.-E., Markus, H.S.: Apathy, but not depression, predicts all-cause dementia in cerebral small vessel disease. Journal of Neurology, Neurosurgery & Psychiatry 91 (9), 953–959 (2020)

Morton, R.E., St. John, P.D., Tyas, S.L.: Migraine and the risk of all-cause dementia, alzheimer’s disease, and vascular dementia: A prospective cohort study in community-dwelling older adults. International journal of geriatric psychiatry 34 (11), 1667–1676 (2019)

Sanford, A.M.: Lewy body dementia. Clinics in geriatric medicine 34 (4), 603–615 (2018)

Taylor, J.-P., McKeith, I.G., Burn, D.J., Boeve, B.F., Weintraub, D., Bamford, C., Allan, L.M., Thomas, A.J., T O’Brien, J.: New evidence on the management of lewy body dementia. The Lancet Neurology 19 (2), 157–169 (2020)

McKEITH, I.G., Galasko, D., Wilcock, G.K., Byrne, E.J.: Lewy body dementia–diagnosis and treatment. The British Journal of Psychiatry 167 (6), 709–717 (1995)

Surendranathan, A., Kane, J.P., Bentley, A., Barker, S.A., Taylor, J.-P., Thomas, A.J., Allan, L.M., McNally, R.J., James, P.W., McKeith, I.G., et al.: Clinical diagnosis of lewy body dementia. BJPsych open 6 (4) (2020)

Finger, E.C.: Frontotemporal dementias. Continuum: Lifelong Learning in Neurology 22 (2 Dementia), 464 (2016)

Rabinovici, G.D., Miller, B.L.: Frontotemporal lobar degeneration. CNS drugs 24 (5), 375–398 (2010)

Organization, W.H., et al.: The ICD-10 classification of mental and behavioural disorders. Clinical descriptions and diagnostic guidelines; 2004. Geneva: WHO (2021)

Ghetti, B., Buratti, E., Boeve, B., Rademakers, R.: Frontotemporal Dementias vol. 320. Springer, (2021)

Custodio, N., Montesinos, R., Lira, D., Herrera-Pérez, E., Bardales, Y., Valeriano-Lorenzo, L.: Mixed dementia: A review of the evidence. Dementia & neuropsychologia 11 , 364–370 (2017)

Garcia-Ptacek, S., Kåreholt, I., Cermakova, P., Rizzuto, D., Religa, D., Eriksdotter, M.: Causes of death according to death certificates in individuals with dementia: a cohort from the swedish dementia registry. Journal of the American Geriatrics Society 64 (11), 137–142 (2016)

Arvanitakis, Z., Shah, R.C., Bennett, D.A.: Diagnosis and management of dementia. Jama 322 (16), 1589–1599 (2019)

Ullah, H., Bin Heyat, M.B., AlSalman, H., Khan, H.M., Akhtar, F., Gumaei, A., Mehdi, A., Muaad, A.Y., Islam, M.S., Ali, A., et al.: An effective and lightweight deep electrocardiography arrhythmia recognition model using novel special and native structural regularization techniques on cardiac signal. Journal of Healthcare Engineering 2022 (2022)

Javeed, A., Ali, L., Mohammed Seid, A., Ali, A., Khan, D., Imrana, Y.: A clinical decision support system (cdss) for unbiased prediction of caesarean section based on features extraction and optimized classification. Computational Intelligence and Neuroscience 2022 (2022)

Imrana, Y., Xiang, Y., Ali, L., Abdul-Rauf, Z., Hu, Y.-C., Kadry, S., Lim, S.: \(\chi\) 2-bidlstm: A feature driven intrusion detection system based on \(\chi\) 2 statistical model and bidirectional lstm. Sensors 22 (5), 2018 (2022)

Ali, L., Zhu, C., Zhao, H., Zhang, Z., Liu, Y.: An integrated system for unbiased parkinson’s disease detection from handwritten drawings. In: Advances in Intelligent Systems and Computing, pp. 3–13. Springer, (2022)

Ali, L., Niamat, A., Khan, J.A., Golilarz, N.A., Xingzhong, X., Noor, A., Nour, R., Bukhari, S.A.C.: An optimized stacked support vector machines based expert system for the effective prediction of heart failure. IEEE Access 7 , 54007–54014 (2019)

Ali, L., Zhu, C., Zhou, M., Liu, Y.: Early diagnosis of parkinson’s disease from multiple voice recordings by simultaneous sample and feature selection. Expert Systems with Applications 137 , 22–28 (2019)

Ali, L., Khan, S.U., Golilarz, N.A., Yakubu, I., Qasim, I., Noor, A., Nour, R.: A feature-driven decision support system for heart failure prediction based on statistical model and gaussian naive bayes. Computational and Mathematical Methods in Medicine 2019 (2019)

Mehbodniya, A., Khan, I.R., Chakraborty, S., Karthik, M., Mehta, K., Ali, L., Nuagah, S.J.: Data mining in employee healthcare detection using intelligence techniques for industry development. Journal of Healthcare Engineering 2022 (2022)

Ali, L., Khan, S.U., Arshad, M., Ali, S., Anwar, M.: A multi-model framework for evaluating type of speech samples having complementary information about parkinson’s disease. In: 2019 International Conference on Electrical, Communication, and Computer Engineering (ICECCE), pp. 1–5 (2019). IEEE

Ahmad, F.S., Ali, L., Khattak, H.A., Hameed, T., Wajahat, I., Kadry, S., Bukhari, S.A.C., et al : A hybrid machine learning framework to predict mortality in paralytic ileus patients using electronic health records (ehrs). Journal of Ambient Intelligence and Humanized Computing 12 (3), 3283–3293 (2021)

Ali, L., Wajahat, I., Golilarz, N.A., Keshtkar, F., Bukhari, S.A.C.: Lda–ga–svm: improved hepatocellular carcinoma prediction through dimensionality reduction and genetically optimized support vector machine. Neural Computing and Applications 33 (7), 2783–2792 (2021)

Ali, L., Zhu, C., Zhang, Z., Liu, Y.: Automated detection of parkinson’s disease based on multiple types of sustained phonations using linear discriminant analysis and genetically optimized neural network. IEEE journal of translational engineering in health and medicine 7 , 1–10 (2019)

Ali, L., He, Z., Cao, W., Rauf, H.T., Imrana, Y., Heyat, M.B.B.: Mmdd-ensemble: A multimodal data–driven ensemble approach for parkinson’s disease detection. Frontiers in Neuroscience 15 (2021)

Rehman, A., Khan, A., Ali, M.A., Khan, M.U., Khan, S.U., Ali, L.: Performance analysis of pca, sparse pca, kernel pca and incremental pca algorithms for heart failure prediction. In: 2020 International Conference on Electrical, Communication, and Computer Engineering (ICECCE), pp. 1–5 (2020). IEEE

Ahmed, F.S., Ali, L., Joseph, B.A., Ikram, A., Mustafa, R.U., Bukhari, S.A.C.: A statistically rigorous deep neural network approach to predict mortality in trauma patients admitted to the intensive care unit. Journal of Trauma and Acute Care Surgery 89 (4), 736–742 (2020)

Imrana, Y., Xiang, Y., Ali, L., Abdul-Rauf, Z.: A bidirectional lstm deep learning approach for intrusion detection. Expert Systems with Applications 185 , 115524 (2021)

Akbar, W., Wu, W.-p., Saleem, S., Farhan, M., Saleem, M.A., Javeed, A., Ali, L.: Development of hepatitis disease detection system by exploiting sparsity in linear support vector machine to improve strength of adaboost ensemble model. Mobile Information Systems 2020 (2020)

Ali, L., Bukhari, S.: An approach based on mutually informed neural networks to optimize the generalization capabilities of decision support systems developed for heart failure prediction. Irbm 42 (5), 345–352 (2021)

McInnes, M.D., Moher, D., Thombs, B.D., McGrath, T.A., Bossuyt, P.M., Clifford, T., Cohen, J.F., Deeks, J.J., Gatsonis, C., Hooft, L., et al : Preferred reporting items for a systematic review and meta-analysis of diagnostic test accuracy studies: the prisma-dta statement. Jama 319 (4), 388–396 (2018)

Liberati, A., Altman, D.G., Tetzlaff, J., Mulrow, C., Gøtzsche, P.C., Ioannidis, J.P., Clarke, M., Devereaux, P.J., Kleijnen, J., Moher, D.: The prisma statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. Journal of clinical epidemiology 62 (10), 1–34 (2009)

Martí-Juan, G., Sanroma-Guell, G., Piella, G.: A survey on machine and statistical learning for longitudinal analysis of neuroimaging data in alzheimer’s disease. Computer methods and programs in biomedicine 189 , 105348 (2020)

McGinnis, J.M., Olsen, L., Goolsby, W.A., Grossmann, C., et al : Clinical Data as the Basic Staple of Health Learning: Creating and Protecting a Public Good: Workshop Summary. National Academies Press, (2011)

Vaughn, V.M., Linder, J.A.: Thoughtless design of the electronic health record drives overuse, but purposeful design can nudge improved patient care. BMJ Publishing Group Ltd (2018)

Doody, R., Stevens, J., Beck, C., Dubinsky, R., Kaye, J., Gwyther, L., Mohs, R., Thal, L., Whitehouse, P., DeKosky, S., et al : Practice parameter: Management of dementia (an evidence-based review): Report of the quality standards subcommittee of the american academy of neurology. Neurology 56 (9), 1154–1166 (2001)

Chi, C.-L., Zeng, W., Oh, W., Borson, S., Lenskaia, T., Shen, X., Tonellato, P.J.: Personalized long-term prediction of cognitive function: Using sequential assessments to improve model performance. Journal of biomedical informatics 76 , 78–86 (2017)

Johnson, K.A., Minoshima, S., Bohnen, N.I., Donohoe, K.J., Foster, N.L., Herscovitch, P., Karlawish, J.H., Rowe, C.C., Carrillo, M.C., Hartley, D.M., et al : Appropriate use criteria for amyloid pet: a report of the amyloid imaging task force, the society of nuclear medicine and molecular imaging, and the alzheimer’s association. Alzheimer’s & Dementia 9 (1), 1–16 (2013)

Shaw, L.M., Arias, J., Blennow, K., Galasko, D., Molinuevo, J.L., Salloway, S., Schindler, S., Carrillo, M.C., Hendrix, J.A., Ross, A., et al : Appropriate use criteria for lumbar puncture and cerebrospinal fluid testing in the diagnosis of alzheimer’s disease. Alzheimer’s & Dementia 14 (11), 1505–1521 (2018)

Mills, K.L., Tamnes, C.K.: Methods and considerations for longitudinal structural brain imaging analysis across development. Developmental cognitive neuroscience 9 , 172–190 (2014)

Ali, L., Zhu, C., Golilarz, N.A., Javeed, A., Zhou, M., Liu, Y.: Reliable parkinson’s disease detection by analyzing handwritten drawings: construction of an unbiased cascaded learning system based on feature selection and adaptive boosting model. Ieee Access 7 , 116480–116489 (2019)

Javeed, A., Rizvi, S.S., Zhou, S., Riaz, R., Khan, S.U., Kwon, S.J.: Heart risk failure prediction using a novel feature selection method for feature refinement and neural network for classification. Mobile Information Systems 2020 (2020)

Javeed, A., Zhou, S., Yongjian, L., Qasim, I., Noor, A., Nour, R.: An intelligent learning system based on random search algorithm and optimized random forest model for improved heart disease detection. IEEE Access 7 , 180235–180243 (2019)

Ali, L., Rahman, A., Khan, A., Zhou, M., Javeed, A., Khan, J.A.: An automated diagnostic system for heart disease prediction based on chi2 statistical model and optimally configured deep neural network. IEEE Access 7 , 34938–34945 (2019)

Maddox, T.M., Rumsfeld, J.S., Payne, P.R.: Questions for artificial intelligence in health care. Jama 321 (1), 31–32 (2019)

Chen, P.-H.C., Liu, Y., Peng, L.: How to develop machine learning models for healthcare. Nature materials 18 (5), 410–414 (2019)

Javeed, A., Khan, S.U., Ali, L., Ali, S., Imrana, Y., Rahman, A.: Machine learning-based automated diagnostic systems developed for heart failure prediction using different types of data modalities: A systematic review and future directions. Computational and Mathematical Methods in Medicine 2022 (2022)

Thacker, S.B., Qualters, J.R., Lee, L.M., for Disease Control, C., Prevention, et al : Public health surveillance in the united states: evolution and challenges. MMWR Suppl 61 (3), 3–9 (2012)

Boerma, J.T., Stansfield, S.K.: Health statistics now: are we making the right investments? The Lancet 369 (9563), 779–786 (2007)

Walport, M., Brest, P.: Sharing research data to improve public health. The Lancet 377 (9765), 537–539 (2011)

Van Panhuis, W.G., Paul, P., Emerson, C., Grefenstette, J., Wilder, R., Herbst, A.J., Heymann, D., Burke, D.S.: A systematic review of barriers to data sharing in public health. BMC public health 14 (1), 1–9 (2014)

Dashtipour, K., Taylor, W., Ansari, S., Zahid, A., Gogate, M., Ahmad, J., Assaleh, K., Arshad, K., Imran, M.A., Abbai, Q.: Detecting alzheimer’s disease using machine learning methods. In: EAI (2021)

Helaly, H.A., Badawy, M., Haikal, A.Y.: Deep learning approach for early detection of alzheimer’s disease. Cognitive Computation, 1–17 (2021)

Vandenberghe, R., Nelissen, N., Salmon, E., Ivanoiu, A., Hasselbalch, S., Andersen, A., Korner, A., Minthon, L., Brooks, D.J., Van Laere, K., et al : Binary classification of 18f-flutemetamol pet using machine learning: comparison with visual reads and structural mri. Neuroimage 64 , 517–525 (2013)

Odusami, M., Maskeliūnas, R., Damaševičius, R., Krilavičius, T.: Analysis of features of alzheimer’s disease: Detection of early stage from functional brain changes in magnetic resonance images using a finetuned resnet18 network. Diagnostics 11 (6), 1071 (2021)

Zheng, Y., Guo, H., Zhang, L., Wu, J., Li, Q., Lv, F.: Machine learning-based framework for differential diagnosis between vascular dementia and alzheimer’s disease using structural mri features. Frontiers in Neurology, 1097 (2019)

Basheer, S., Bhatia, S., Sakri, S.B.: Computational modeling of dementia prediction using deep neural network: Analysis on oasis dataset. IEEE Access 9 , 42449–42462 (2021)

Leong, L.K., Abdullah, A.A.: Prediction of alzheimer’s disease (ad) using machine learning techniques with boruta algorithm as feature selection method. In: Journal of Physics: Conference Series, vol. 1372, p. 012065 (2019). IOP Publishing

Battineni, G., Chintalapudi, N., Amenta, F.: Machine learning in medicine: Performance calculation of dementia prediction by support vector machines (svm). Informatics in Medicine Unlocked 16 , 100200 (2019)

Mathotaarachchi, S., Pascoal, T.A., Shin, M., Benedet, A.L., Kang, M.S., Beaudry, T., Fonov, V.S., Gauthier, S., Rosa-Neto, P., Initiative, A.D.N., et al : Identifying incipient dementia individuals using machine learning and amyloid imaging. Neurobiology of aging 59 , 80–90 (2017)

Aruna, S., Chitra, S.: Machine learning approach for identifying dementia from mri images. International Journal of Computer and Information Engineering 9 (3), 881–888 (2016)

Herzog, N.J., Magoulas, G.D.: Brain asymmetry detection and machine learning classification for diagnosis of early dementia. Sensors 21 (3), 778 (2021)

Battineni, G., Chintalapudi, N., Amenta, F., Traini, E.: A comprehensive machine-learning model applied to magnetic resonance imaging (mri) to predict alzheimer’s disease (ad) in older subjects. Journal of Clinical Medicine 9 (7), 2146 (2020)

Ma, D., Yee, E., Stocks, J.K., Jenkins, L.M., Popuri, K., Chausse, G., Wang, L., Probst, S., Beg, M.F.: Blinded clinical evaluation for dementia of alzheimer’s type classification using fdg-pet: A comparison between feature-engineered and non-feature-engineered machine learning methods. Journal of Alzheimer’s Disease 80 (2), 715–726 (2021)

Bidani, A., Gouider, M.S., Travieso-González, C.M.: Dementia detection and classification from mri images using deep neural networks and transfer learning. In: International Work-Conference on Artificial Neural Networks, pp. 925–933 (2019). Springer

Moscoso, A., Silva-Rodríguez, J., Aldrey, J.M., Cortés, J., Fernández-Ferreiro, A., Gómez-Lado, N., Ruibal, Á., Aguiar, P., Initiative, A.D.N., et al : Prediction of alzheimer’s disease dementia with mri beyond the short-term: Implications for the design of predictive models. NeuroImage: Clinical 23 , 101837 (2019)

Khan, A., Zubair, S.: An improved multi-modal based machine learning approach for the prognosis of alzheimer’s disease. Journal of King Saud University-Computer and Information Sciences (2020)

Mohammed, B.A., Senan, E.M., Rassem, T.H., Makbol, N.M., Alanazi, A.A., Al-Mekhlafi, Z.G., Almurayziq, T.S., Ghaleb, F.A.: Multi-method analysis of medical records and mri images for early diagnosis of dementia and alzheimer’s disease based on deep learning and hybrid methods. Electronics 10 (22), 2860 (2021)

Salvatore, C., Cerasa, A., Battista, P., Gilardi, M.C., Quattrone, A., Castiglioni, I.: Magnetic resonance imaging biomarkers for the early diagnosis of alzheimer’s disease: a machine learning approach. Frontiers in neuroscience 9 , 307 (2015)

Katako, A., Shelton, P., Goertzen, A.L., Levin, D., Bybel, B., Aljuaid, M., Yoon, H.J., Kang, D.Y., Kim, S.M., Lee, C.S., et al : Machine learning identified an alzheimer’s disease-related fdg-pet pattern which is also expressed in lewy body dementia and parkinson’s disease dementia. Scientific reports 8 (1), 1–13 (2018)

Gray, K.R., Aljabar, P., Heckemann, R.A., Hammers, A., Rueckert, D.: Random forest-based manifold learning for classification of imaging data in dementia. In: International Workshop on Machine Learning in Medical Imaging, pp. 159–166 (2011). Springer

De Bruijne, M.: Machine learning approaches in medical image analysis: From detection to diagnosis. Elsevier (2016)

Tong, T., Wolz, R., Gao, Q., Hajnal, J.V., Rueckert, D.: Multiple instance learning for classification of dementia in brain mri. In: International Conference on Medical Image Computing and Computer-Assisted Intervention, pp. 599–606 (2013). Springer

Akhila, J., Markose, C., Aneesh, R.: Feature extraction and classification of dementia with neural network. In: 2017 International Conference on Intelligent Computing, Instrumentation and Control Technologies (ICICICT), pp. 1446–1450 (2017). IEEE

Chen, Y., Pham, T.D.: Development of a brain mri-based hidden markov model for dementia recognition. Biomedical engineering online 12 (1), 1–16 (2013)

Patil, M., Yardi, A.: Ann based dementia diagnosis using dct for brain mr image compression. In: 2013 International Conference on Communication and Signal Processing, pp. 451–454 (2013). IEEE

Gulhare, K.K., Shukla, S., Sharma, L.: Deep neural network classification method to alzheimer’s disease detection. International Journals of Advanced Research in Computer Science and Software Engineering 7 (6), 1–4 (2017)

Chiu, P.-Y., Tang, H., Wei, C.-Y., Zhang, C., Hung, G.-U., Zhou, W.: Nmd-12: A new machine-learning derived screening instrument to detect mild cognitive impairment and dementia. PloS one 14 (3), 0213430 (2019)

Stamate, D., Alghamdi, W., Ogg, J., Hoile, R., Murtagh, F.: A machine learning framework for predicting dementia and mild cognitive impairment. In: 2018 17th IEEE International Conference on Machine Learning and Applications (ICMLA), pp. 671–678 (2018). IEEE

Visser, P.J., Lovestone, S., Legido-Quigley, C.: A metabolite-based machine learning approach to diagnose alzheimer-type dementia in blood: Results from the european medical information framework for alzheimer disease biomarker discovery cohort (2019)

Bansal, D., Chhikara, R., Khanna, K., Gupta, P.: Comparative analysis of various machine learning algorithms for detecting dementia. Procedia computer science 132 , 1497–1502 (2018)

Nori, V.S., Hane, C.A., Martin, D.C., Kravetz, A.D., Sanghavi, D.M.: Identifying incident dementia by applying machine learning to a very large administrative claims dataset. PLoS One 14 (7), 0203246 (2019)

Alam, M.A.U., Roy, N., Holmes, S., Gangopadhyay, A., Galik, E.: Automated functional and behavioral health assessment of older adults with dementia. In: 2016 IEEE First International Conference on Connected Health: Applications, Systems and Engineering Technologies (CHASE), pp. 140–149 (2016). IEEE

Gurevich, P., Stuke, H., Kastrup, A., Stuke, H., Hildebrandt, H.: Neuropsychological testing and machine learning distinguish alzheimer’s disease from other causes for cognitive impairment. Frontiers in aging neuroscience 9 , 114 (2017)

Karaglani, M., Gourlia, K., Tsamardinos, I., Chatzaki, E.: Accurate blood-based diagnostic biosignatures for alzheimer’s disease via automated machine learning. Journal of clinical medicine 9 (9), 3016 (2020)

Ryzhikova, E., Ralbovsky, N.M., Sikirzhytski, V., Kazakov, O., Halamkova, L., Quinn, J., Zimmerman, E.A., Lednev, I.K.: Raman spectroscopy and machine learning for biomedical applications: Alzheimer’s disease diagnosis based on the analysis of cerebrospinal fluid. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 248 , 119188 (2021)

Cho, P.-C., Chen, W.-H.: A double layer dementia diagnosis system using machine learning techniques. In: International Conference on Engineering Applications of Neural Networks, pp. 402–412 (2012). Springer

Facal, D., Valladares-Rodriguez, S., Lojo-Seoane, C., Pereiro, A.X., Anido-Rifon, L., Juncos-Rabadán, O.: Machine learning approaches to studying the role of cognitive reserve in conversion from mild cognitive impairment to dementia. International journal of geriatric psychiatry 34 (7), 941–949 (2019)

Jin, H., Chien, S., Meijer, E., Khobragade, P., Lee, J., et al : Learning from clinical consensus diagnosis in india to facilitate automatic classification of dementia: Machine learning study. JMIR Mental Health 8 (5), 27113 (2021)

James, C., Ranson, J.M., Everson, R., Llewellyn, D.J.: Performance of machine learning algorithms for predicting progression to dementia in memory clinic patients. JAMA network open 4 (12), 2136553–2136553 (2021)

Bougea, A., Efthymiopoulou, E., Spanou, I., Zikos, P.: A novel machine learning algorithm predicts dementia with lewy bodies versus parkinson’s disease dementia based on clinical and neuropsychological scores. Journal of geriatric psychiatry and neurology, 0891988721993556 (2021)

Salem, F.A., Chaaya, M., Ghannam, H., Al Feel, R.E., El Asmar, K.: Regression based machine learning model for dementia diagnosis in a community setting. Alzheimer’s & Dementia 17 , 053839 (2021)

Garcia-Gutierrez, F., Delgado-Alvarez, A., Delgado-Alonso, C., Díaz-Álvarez, J., Pytel, V., Valles-Salgado, M., Gil, M.J., Hernández-Lorenzo, L., Matías-Guiu, J., Ayala, J.L., et al.: Diagnosis of alzheimer’s disease and behavioural variant frontotemporal dementia with machine learning-aided neuropsychological assessment using feature engineering and genetic algorithms. International journal of geriatric psychiatry 37 (2) (2022)

Mirzaei, G., Adeli, H.: Machine learning techniques for diagnosis of alzheimer disease, mild cognitive disorder, and other types of dementia. Biomedical Signal Processing and Control 72 , 103293 (2022)

Hsiu, H., Lin, S.-K., Weng, W.-L., Hung, C.-M., Chang, C.-K., Lee, C.-C., Chen, C.-T.: Discrimination of the cognitive function of community subjects using the arterial pulse spectrum and machine-learning analysis. Sensors 22 (3), 806 (2022)

Shahzad, A., Dadlani, A., Lee, H., Kim, K.: Automated prescreening of mild cognitive impairment using shank-mounted inertial sensors based gait biomarkers. IEEE Access (2022)

Hane, C.A., Nori, V.S., Crown, W.H., Sanghavi, D.M., Bleicher, P.: Predicting onset of dementia using clinical notes and machine learning: case-control study. JMIR medical informatics 8 (6), 17819 (2020)

Aschwanden, D., Aichele, S., Ghisletta, P., Terracciano, A., Kliegel, M., Sutin, A.R., Brown, J., Allemand, M.: Predicting cognitive impairment and dementia: A machine learning approach. Journal of Alzheimer’s Disease 75 (3), 717–728 (2020)

Ryu, S.-E., Shin, D.-H., Chung, K.: Prediction model of dementia risk based on xgboost using derived variable extraction and hyper parameter optimization. IEEE Access 8 , 177708–177720 (2020)

de Langavant, L.C., Bayen, E., Yaffe, K., et al : Unsupervised machine learning to identify high likelihood of dementia in population-based surveys: development and validation study. Journal of medical Internet research 20 (7), 10493 (2018)

Fouladvand, S., Mielke, M.M., Vassilaki, M., Sauver, J.S., Petersen, R.C., Sohn, S.: Deep learning prediction of mild cognitive impairment using electronic health records. In: 2019 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), pp. 799–806 (2019). IEEE

Balea-Fernandez, F.J., Martinez-Vega, B., Ortega, S., Fabelo, H., Leon, R., Callico, G.M., Bibao-Sieyro, C.: Analysis of risk factors in dementia through machine learning. Journal of Alzheimer’s Disease 79 (2), 845–861 (2021)

Chlasta, K., Wołk, K.: Towards computer-based automated screening of dementia through spontaneous speech. Frontiers in Psychology, 4091 (2021)

Chien, Y.-W., Hong, S.-Y., Cheah, W.-T., Yao, L.-H., Chang, Y.-L., Fu, L.-C.: An automatic assessment system for alzheimer’s disease based on speech using feature sequence generator and recurrent neural network. Scientific Reports 9 (1), 1–10 (2019)

Shimoda, A., Li, Y., Hayashi, H., Kondo, N.: Dementia risks identified by vocal features via telephone conversations: A novel machine learning prediction model. PloS one 16 (7), 0253988 (2021)

Nishikawa, K., Akihiro, K., Hirakawa, R., Kawano, H., Nakatoh, Y.: Machine learning model for discrimination of mild dementia patients using acoustic features. Cognitive Robotics (2021)

Liu, L., Zhao, S., Chen, H., Wang, A.: A new machine learning method for identifying alzheimer’s disease. Simulation Modelling Practice and Theory 99 , 102023 (2020)

Searle, T., Ibrahim, Z., Dobson, R.: Comparing natural language processing techniques for alzheimer’s dementia prediction in spontaneous speech. arXiv preprint arXiv:2006.07358 (2020)

Zhu, Y., Tran, B., Liang, X., Batsis, J.A., Roth, R.M.: Towards interpretability of speech pause in dementia detection using adversarial learning. arXiv preprint arXiv:2111.07454 (2021)

Ossewaarde, R., Jonkers, R., Jalvingh, F., Bastiaanse, R.: Classification of spontaneous speech of individuals with dementia based on automatic prosody analysis using support vector machines (svm). In: The Thirty-Second International Flairs Conference (2019)

Xue, C., Karjadi, C., Paschalidis, I.C., Au, R., Kolachalama, V.B.: Detection of dementia on voice recordings using deep learning: a framingham heart study. Alzheimer’s research & therapy 13 (1), 1–15 (2021)

Weiner, J., Engelbart, M., Schultz, T.: Manual and automatic transcriptions in dementia detection from speech. In: INTERSPEECH, pp. 3117–3121 (2017)

Sadeghian, R., Schaffer, J.D., Zahorian, S.A.: Speech processing approach for diagnosing dementia in an early stage (2017)

Khodabakhsh, A., Yesil, F., Guner, E., Demiroglu, C.: Evaluation of linguistic and prosodic features for detection of alzheimer’s disease in turkish conversational speech. EURASIP Journal on Audio, Speech, and Music Processing 2015 (1), 1–15 (2015)

Edwards, E., Dognin, C., Bollepalli, B., Singh, M.K., Analytics, V.: Multiscale system for alzheimer’s dementia recognition through spontaneous speech. In: INTERSPEECH, pp. 2197–2201 (2020)

Kumar, Y., Maheshwari, P., Joshi, S., Baths, V.: Ml-based analysis to identify speech features relevant in predicting alzheimer’s disease. arXiv preprint arXiv:2110.13023 (2021)

Luz, S., de la Fuente, S., Albert, P.: A method for analysis of patient speech in dialogue for dementia detection. arXiv preprint arXiv:1811.09919 (2018)

Syed, Z.S., Syed, M.S.S., Lech, M., Pirogova, E.: Automated recognition of alzheimer’s dementia using bag-of-deep-features and model ensembling. IEEE Access 9 , 88377–88390 (2021)

Orsulic-Jeras, S., Sanders, B., Powers, S., Ejaz, F., Cordell, A., Wilk, C.: Developing a reminiscence therapy platform-lifebio memorytm-with a novel machine-learning-based application that transfers speech to text and generates life stories for nursing home residents with dementia. Alzheimer’s & Dementia 17 , 052281 (2021)

Sarawgi, U., Zulfikar, W., Soliman, N., Maes, P.: Multimodal inductive transfer learning for detection of alzheimer’s dementia and its severity. arXiv preprint arXiv:2009.00700 (2020)

Calzà, L., Gagliardi, G., Favretti, R.R., Tamburini, F.: Linguistic features and automatic classifiers for identifying mild cognitive impairment and dementia. Computer Speech & Language 65 , 101113 (2021)

Haider, F., De La Fuente, S., Luz, S.: An assessment of paralinguistic acoustic features for detection of alzheimer’s dementia in spontaneous speech. IEEE Journal of Selected Topics in Signal Processing 14 (2), 272–281 (2019)

Lopez-de-Ipiña, K., Alonso, J.B., Solé-Casals, J., Barroso, N., Henriquez, P., Faundez-Zanuy, M., Travieso, C.M., Ecay-Torres, M., Martinez-Lage, P., Eguiraun, H.: On automatic diagnosis of alzheimer’s disease based on spontaneous speech analysis and emotional temperature. Cognitive Computation 7 (1), 44–55 (2015)

Orimaye, S.O., Wong, J.S.-M., Golden, K.J.: Learning predictive linguistic features for alzheimer’s disease and related dementias using verbal utterances. In: Proceedings of the Workshop on Computational Linguistics and Clinical Psychology: From Linguistic Signal to Clinical Reality, pp. 78–87 (2014)

Santander-Cruz, Y., Salazar-Colores, S., Paredes-García, W.J., Guendulain-Arenas, H., Tovar-Arriaga, S.: Semantic feature extraction using sbert for dementia detection. Brain Sciences 12 (2), 270 (2022)

Download references

Acknowledgements

The first author’s learning process was supported by the National E-Infrastructure for Aging Research (NEAR), Sweden. NEAR is working on improving the health condition of older adults in Sweden.

Open access funding provided by Blekinge Institute of Technology. This research received no external funding.

Author information

Ashir Javeed and Ana Luiza Dallora contributed equally to this work.

Authors and Affiliations

Aging Research Center, Karolinska Institutet, Tomtebodavagen, Stockholm, 17165, Solna, Sweden

Ashir Javeed

Department of Health, Blekinge Institute of Technology, Valhallavägen 1, Karlskrona, 37141, Blekinge, Sweden

Ashir Javeed, Ana Luiza Dallora, Johan Sanmartin Berglund & Peter Anderberg

Department of Computer Science, University of Science and Technology Bannu, Township, Bannu, 28100, Khyber-Pakhtunkhwa, Pakistan

Department of Electrical Engineering, University of Science and Technology Bannu, Township, Bannu, 28100, Khyber-Pakhtunkhwa, Pakistan

Liaqata Ali

School of Health Sciences, University of Skovde, Högskolevägen 1, Skövde, SE-541 28, Skövde, Sweden

Peter Anderberg

You can also search for this author in PubMed   Google Scholar

Contributions

Conceptualization by Peter Anderber, Data curation by Liaqat ALi, Formal analysis by Ana Luiza Dallora, Write up and Methodology by Ashir Javeed, Proofread by Arif Ali, Supervised by Johan Sanmartin Berglund. If any of the sections are not relevant to your manuscript.

Corresponding author

Correspondence to Johan Sanmartin Berglund .

Ethics declarations

Ethics approval.

This study was carried out in accordance with the Declaration of Helsinki and was approved by the Research Ethics Committee at Blekinge Institute of Technology (BTH).

Consent to Participate

Consent for publication, conflict of interest.

The authors declare no conflict of interest.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Javeed, A., Dallora, A.L., Berglund, J.S. et al. Machine Learning for Dementia Prediction: A Systematic Review and Future Research Directions. J Med Syst 47 , 17 (2023). https://doi.org/10.1007/s10916-023-01906-7

Download citation

Received : 16 July 2022

Accepted : 03 January 2023

Published : 01 February 2023

DOI : https://doi.org/10.1007/s10916-023-01906-7

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Dementia prediction
• Feature selection
• Machine learning
• Deep learning
• Find a journal
• Publish with us
• Track your research

• Introduction
• Conclusions
• Article Information

Models were stratified by age, cohort (sex), and calendar time, and adjusted for Southern European/Mediterranean ancestry (yes/no), married (yes/no), living alone (yes/no), smoking status (never, former, current smoker 1-14 cigarettes/d, 15-24 cigarettes/d, or ≥25 cigarettes/d), physical activity (<3.0, 3.0-8.9, 9.0-17.9, 18.0-26.9, ≥27.0 metabolic equivalent of task–h/wk), multivitamin use (yes/no), history of hypertension (yes/no), history of hypercholesterolemia (yes/no), history of diabetes (yes/no), in women postmenopausal status and menopausal hormone use (premenopausal, postmenopausal [no, past, or current hormone use]), total energy intake (kcal/d), family history of dementia (yes/no), history of depression (yes/no), census socioeconomic status (9-variable score, in quintiles), and body mass index calculated as weight in kilograms divided by height in meters squared (<23, 23-25, 25-30, 30-35, ≥35). Pooled results were obtained by pooling the datasets of the cohorts. AMED score is without monounsaturated:saturated fats intake ratio component. AHEI score is without polyunsaturated fats intake component. HR indicates hazard ratio.

a Reference value.

b P  < .05.

Substitution analysis of 5 g/d intake of olive oil for the equivalent amount of butter, other vegetable oils, mayonnaise, and margarine. All Cox proportional hazards models were stratified by age and calendar time. Models were adjusted for Southern European/Mediterranean ancestry (yes/no), married (yes/no), living alone (yes/no), smoking status (never, former, current smoker 1-14 cigarettes/d, 15-24 cigarettes/d, or ≥25 cigarettes/d), alcohol intake (0, 0.1-4.9, 5.0-9.9, 10.0-14.9, and ≥15.0 g/d), physical activity (<3.0, 3.0-8.9, 9.0-17.9, 18.0-26.9, ≥27.0 metabolic equivalent of task–h/wk), multivitamin use (yes/no), history of hypertension (yes/no), history of hypercholesterolemia (yes/no), in women postmenopausal status and menopausal hormone use (premenopausal, postmenopausal [no, past, or current hormone use]), total energy intake (kcal/d), family history of dementia (yes/no), history of depression (yes/no), census socioeconomic status (9-variable score, in quintiles), body mass index calculated as weight in kilograms divided by height in meters squared (<23, 23-25, 25-30, 30-35, ≥35), red meat, fruits and vegetables, nuts, soda, whole grains intake (in quintiles), and trans-fat. Pooled results were obtained by pooling the data sets of the cohorts and Cox proportional hazards model 3 was further stratified by cohort (sex). HR indicates hazard ratio.

eTable 1. Odds Ratios for Dementia-Related Mortality by APOE4 Allelic Dosage

eTable 2. Risk of Death With Dementia (Composite Outcome) According to Categories of Total Olive Oil

eTable 3. Joint Associations of Olive Oil Intake and AMED (A), and AHEI (B) With Dementia-Related Mortality Risk

eTable 4. Risk of Dementia-Related Mortality According to Categories of Total Olive Oil in the Genomic DNA Subsample

eFigure. Subgroup Analyses for 5g/d Increase in Olive Oil Intake With Dementia-Related Mortality Risk

eTable 5. Risk of Dementia-Related Mortality According to Categories of Total Olive Oil Without Stopping Diet Update Upon Report of Intermediate Non-Fatal Events

eTable 6. Risk of Dementia Mortality According to Categories of Total Olive Oil Applying a 4-Year Lag Period Between Dietary Intake and Dementia Mortality

eTable 7. Risk of Dementia-Related Mortality According to Categories of Total Olive Oil Adjusting for Other Covariates

eTable 8. Risk of Mortality From Dementia and Other Causes of Death According to Categories of Total Olive Oil Applying a Competing Risk Model

eReferences

Data Sharing Statement

See More About

Sign up for emails based on your interests, select your interests.

Customize your JAMA Network experience by selecting one or more topics from the list below.

• Academic Medicine
• Acid Base, Electrolytes, Fluids
• Allergy and Clinical Immunology
• American Indian or Alaska Natives
• Anesthesiology
• Anticoagulation
• Art and Images in Psychiatry
• Artificial Intelligence
• Assisted Reproduction
• Bleeding and Transfusion
• Caring for the Critically Ill Patient
• Challenges in Clinical Electrocardiography
• Climate and Health
• Climate Change
• Clinical Challenge
• Clinical Decision Support
• Clinical Implications of Basic Neuroscience
• Clinical Pharmacy and Pharmacology
• Complementary and Alternative Medicine
• Consensus Statements
• Coronavirus (COVID-19)
• Critical Care Medicine
• Cultural Competency
• Dental Medicine
• Dermatology
• Diabetes and Endocrinology
• Diagnostic Test Interpretation
• Drug Development
• Electronic Health Records
• Emergency Medicine
• End of Life, Hospice, Palliative Care
• Environmental Health
• Equity, Diversity, and Inclusion
• Facial Plastic Surgery
• Gastroenterology and Hepatology
• Genetics and Genomics
• Genomics and Precision Health
• Global Health
• Guide to Statistics and Methods
• Hair Disorders
• Health Care Delivery Models
• Health Care Economics, Insurance, Payment
• Health Care Quality
• Health Care Reform
• Health Care Safety
• Health Care Workforce
• Health Disparities
• Health Inequities
• Health Policy
• Health Systems Science
• History of Medicine
• Hypertension
• Images in Neurology
• Implementation Science
• Infectious Diseases
• Innovations in Health Care Delivery
• JAMA Infographic
• Law and Medicine
• Leading Change
• Less is More
• LGBTQIA Medicine
• Lifestyle Behaviors
• Medical Coding
• Medical Devices and Equipment
• Medical Education
• Medical Education and Training
• Medical Journals and Publishing
• Mobile Health and Telemedicine
• Narrative Medicine
• Neuroscience and Psychiatry
• Notable Notes
• Nutrition, Obesity, Exercise
• Obstetrics and Gynecology
• Occupational Health
• Ophthalmology
• Orthopedics
• Otolaryngology
• Pain Medicine
• Palliative Care
• Pathology and Laboratory Medicine
• Patient Care
• Patient Information
• Performance Improvement
• Performance Measures
• Perioperative Care and Consultation
• Pharmacoeconomics
• Pharmacoepidemiology
• Pharmacogenetics
• Pharmacy and Clinical Pharmacology
• Physical Medicine and Rehabilitation
• Physical Therapy
• Physician Leadership
• Population Health
• Primary Care
• Professional Well-being
• Professionalism
• Psychiatry and Behavioral Health
• Public Health
• Pulmonary Medicine
• Regulatory Agencies
• Reproductive Health
• Research, Methods, Statistics
• Resuscitation
• Rheumatology
• Risk Management
• Scientific Discovery and the Future of Medicine
• Shared Decision Making and Communication
• Sleep Medicine
• Sports Medicine
• Stem Cell Transplantation
• Substance Use and Addiction Medicine
• Surgical Innovation
• Surgical Pearls
• Teachable Moment
• Technology and Finance
• The Art of JAMA
• The Arts and Medicine
• The Rational Clinical Examination
• Tobacco and e-Cigarettes
• Translational Medicine
• Trauma and Injury
• Treatment Adherence
• Ultrasonography
• Users' Guide to the Medical Literature
• Vaccination
• Venous Thromboembolism
• Veterans Health
• Women's Health
• Workflow and Process
• Wound Care, Infection, Healing

Get the latest research based on your areas of interest.

Others also liked.

• Download PDF
• X Facebook More LinkedIn

Tessier A , Cortese M , Yuan C, et al. Consumption of Olive Oil and Diet Quality and Risk of Dementia-Related Death. JAMA Netw Open. 2024;7(5):e2410021. doi:10.1001/jamanetworkopen.2024.10021

Manage citations:

© 2024

• Permissions

Consumption of Olive Oil and Diet Quality and Risk of Dementia-Related Death

• 1 Department of Nutrition, Harvard T.H. Chan School of Public Health, Boston, Massachusetts
• 2 School of Public Health, the Second Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China
• 3 Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, Massachusetts
• 4 Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts
• 5 Department of Public Health and Novo Nordisk Foundation Center for Basic Metabolic Research, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark

Question   Is the long-term consumption of olive oil associated with dementia-related death risk?

Findings   In a prospective cohort study of 92 383 adults observed over 28 years, the consumption of more than 7 g/d of olive oil was associated with a 28% lower risk of dementia-related death compared with never or rarely consuming olive oil, irrespective of diet quality.

Meaning   These results suggest that olive oil intake represents a potential strategy to reduce dementia mortality risk.

Importance   Age-standardized dementia mortality rates are on the rise. Whether long-term consumption of olive oil and diet quality are associated with dementia-related death is unknown.

Objective   To examine the association of olive oil intake with the subsequent risk of dementia-related death and assess the joint association with diet quality and substitution for other fats.

Design, Setting, and Participants   This prospective cohort study examined data from the Nurses’ Health Study (NHS; 1990-2018) and Health Professionals Follow-Up Study (HPFS; 1990-2018). The population included women from the NHS and men from the HPFS who were free of cardiovascular disease and cancer at baseline. Data were analyzed from May 2022 to July 2023.

Exposures   Olive oil intake was assessed every 4 years using a food frequency questionnaire and categorized as (1) never or less than once per month, (2) greater than 0 to less than or equal to 4.5 g/d, (3) greater than 4.5 g/d to less than or equal to 7 g/d, and (4) greater than 7 g/d. Diet quality was based on the Alternative Healthy Eating Index and Mediterranean Diet score.

Main Outcome and Measure   Dementia death was ascertained from death records. Multivariable Cox proportional hazards regressions were used to estimate hazard ratios (HRs) and 95% CIs adjusted for confounders including genetic, sociodemographic, and lifestyle factors.

Results   Of 92 383 participants, 60 582 (65.6%) were women and the mean (SD) age was 56.4 (8.0) years. During 28 years of follow-up (2 183 095 person-years), 4751 dementia-related deaths occurred. Individuals who were homozygous for the apolipoprotein ε4 ( APOE ε4 ) allele were 5 to 9 times more likely to die with dementia. Consuming at least 7 g/d of olive oil was associated with a 28% lower risk of dementia-related death (adjusted pooled HR, 0.72 [95% CI, 0.64-0.81]) compared with never or rarely consuming olive oil ( P for trend < .001); results were consistent after further adjustment for APOE ε4 . No interaction by diet quality scores was found. In modeled substitution analyses, replacing 5 g/d of margarine and mayonnaise with the equivalent amount of olive oil was associated with an 8% (95% CI, 4%-12%) to 14% (95% CI, 7%-20%) lower risk of dementia mortality. Substitutions for other vegetable oils or butter were not significant.

Conclusions and Relevance   In US adults, higher olive oil intake was associated with a lower risk of dementia-related mortality, irrespective of diet quality. Beyond heart health, the findings extend the current dietary recommendations of choosing olive oil and other vegetable oils for cognitive-related health.

One-third of older adults die with Alzheimer disease or another dementia. 1 While deaths from diseases such as stroke and heart disease have been decreasing over the past 20 years, age-standardized dementia mortality rates have been on the rise. 2 The Mediterranean diet has gained in popularity owing to its recognized, multifaceted health benefits, particularly on cardiovascular outcomes. 3 Accruing evidence suggests this dietary pattern also has a beneficial effect on cognitive health. 4 As part of the Mediterranean diet, olive oil may exert anti-inflammatory and neuroprotective effects due to its high content of monounsaturated fatty acids and other compounds with antioxidant properties such as vitamin E and polyphenols. 5 A substudy conducted as part of the Prevencion con Dieta Mediterranea (PREDIMED) randomized trial provided evidence that higher intake of olive oil for 6.5 years combined with adherence to a Mediterranean diet was protective of cognitive decline when compared with a low-fat control diet. 6 - 8

Given that most previous studies on olive oil consumption and cognition were conducted in Mediterranean countries, 7 - 10 studying the US population, where olive oil consumption is generally lower, could offer unique insights. Recently, we showed that olive oil consumption was associated with a lower risk of total and cause-specific mortality in large US prospective cohort studies, including a 29% (95% CI, 22%-36%) lower risk for neurodegenerative disease mortality in participants who consumed more than 7 g/d of olive oil compared with little or none. 11 However, this previous analysis was not designed to examine the association of olive oil and diet quality with dementia-related mortality, and therefore the latter remains unclear.

In this study, we examined the association between total olive oil consumption and the subsequent risk of dementia-related mortality in 2 large prospective studies of US women and men. Additionally, we evaluated the joint associations of diet quality (adherence to the Mediterranean diet and Alternative Healthy Eating Index [AHEI] score) and olive oil consumption with the risk of dementia-related mortality. We also estimated the difference in the risk of dementia-related mortality when other dietary fats were substituted with an equivalent amount of olive oil.

Analyses were performed in 2 large US prospective cohorts: the Nurses’ Health Study I (NHS) and the Health Professionals Follow-Up Study (HPFS). The NHS was initiated in 1976 and recruited 121 700 US female registered nurses aged 30 to 55 years. 12 The HPFS was established in 1986 and included 51 525 male health professionals aged 40 to 75 years. 13 The cohorts have been described elsewhere. 12 , 13 Lifestyle factors and medical history were assessed biennially through mailed questionnaires, with a follow-up rate greater than 90%. Baseline for this analysis was 1990, which is when the food frequency questionnaires (FFQs) first included information on olive oil consumption.

Participants with a history of cardiovascular disease (CVD) or cancer at baseline, with missing data on olive oil consumption, or who reported implausible total energy intakes (<500 or >3500 kcal/d for women and <800 or >4200 kcal/d for men) were excluded. The completion of the questionnaire self-selected cognitively highly functioning women and men. In total, 60 582 women and 31 801 men were included. The study protocol was approved by the institutional review boards of the Brigham and Women’s Hospital and Harvard T.H. Chan School of Public Health, which deemed the participants’ completion of the questionnaire to be considered as implied consent. This report followed the Strengthening the Reporting of Observational Studies in Epidemiology ( STROBE ) reporting guideline.

Dietary intake was measured using a validated greater than 130-item FFQ administered in 1990 and every 4 years thereafter. The validity and reliability of the FFQ have been described previously. 14 Participants were asked how frequently they consumed specific foods, including types of fats and oils used for cooking or added to meals in the past 12 months. Total olive oil intake was determined by summing up answers to 3 questions related to olive oil consumption (ie, olive oil used for salad dressings, olive oil added to food or bread, and olive oil used for baking and frying at home). The equivalent of 1 tablespoon of olive oil was considered to be 13.5 g. Intakes of other fats and nutrients were calculated using the United States Department of Agriculture and Harvard University Food Composition Database, 15 and biochemical analyses. The nutritional composition of olive oil and other types of fat, as well as trends of types of fat intake in the NHS and HPFS, have been reported previously. 11

Adherence to the Mediterranean diet was assessed using a modified version of the 9-point Alternative Mediterranean index (AMED) score. 16 Adherence to the AHEI (0-110), previously associated with lower risk of chronic disease, was also computed. 17 Higher scores indicated better overall diet quality.

The apolipoprotein E ε4 ( APOE ε4 ) allele is known to interfere with lipid and glucose metabolism such that it increases the risk of dementia. 18  APOE genotyping was conducted in a subset of 27 296 participants. Blood samples were collected between 1989 and 1990 in the NHS and between 1993 and 1995 in the HPFS. NHS participants who had not provided blood samples were invited to contribute buccal samples from 2002 to 2004. DNA was extracted with the ReturPureGene DNA Isolation Kit (Gentra Systems). The APOE genotype was determined using a Taqman Assay (Applied Biosystems) 19 in 5069 participants, and through imputation from multiple genome-wide association studies, 20 which has shown high accuracy, 20 in the remaining subset.

Deaths were ascertained from state vital statistics records and the National Death Index or by reports from next of kin or the postal authorities. The follow-up for mortality exceeded 98% in these cohorts. Dementia deaths were determined by physician review of medical records, autopsy reports, or death certificates. Dementia deaths were those in which dementia was listed as the underlying cause of death, or as a contributing cause of death, or as reported by the family, in the absence of a more likely cause. The International Classification of Diseases, Eighth Revision (ICD-8) was used in the NHS and ICD-9 in the HPFS, which were the revisions used at the inception of those cohorts. Dementia deaths included codes 290.0 (senile dementia, simple type), 290.1 (presenile dementia), and 331.0 (Alzheimer disease). To test the validity of the dementia mortality outcome, we examined the likelihood of dementia mortality by APOE ε4 allelic dosage (eTable 1 in Supplement 1 ). 18 A composite outcome was also created including both participants who reported having dementia during follow-up and later died, with those who had dementia reported on their death certificate.

Participants completed biennial questionnaires reporting updates on body weight, smoking, physical activity, multivitamin use, menopausal status, and postmenopausal hormone use in women, family history of dementia, self-report of chronic diseases, and ancestry. History of depression was identified based on antidepressive medication use and self-report of depression. Socioeconomic status (SES) was established through a composite score derived from home address details and various factors such as income, education, and housing; the composite score methods are described in a previous report. 21 Body mass index (BMI) was obtained by dividing the weight in kilograms by the height in meters squared.

In each cohort, age-stratified Cox proportional hazard models were used to evaluate the association of olive oil intake with dementia-related mortality. Participant person-time was calculated from baseline until end of follow-up (June 30, 2018, in NHS; January 31, 2018, in HPFS), loss to follow-up, or death, whichever came first. The cumulative average (mean) of olive oil intake from all available FFQs, from baseline until 2014 (or loss to follow-up or death), was used as the exposure. Because potential diet modifications following cancer or CVD diagnosis may not represent long-term diet, we ceased updating dietary variables upon report of these conditions. For missing covariates, we carried forward nonmissing values from previous questionnaires and assigned median values for continuous variables.

Participants were categorized by olive oil intake frequency: never or less than once per month (reference group), greater than 0 to less than or equal to 4.5 g/d, greater than 4.5 g/d to less than or equal to 7 g/d, and greater than 7 g/d. P values for linear trends were obtained using the Wald test on a continuous variable represented by the median intake of each category. Multivariable Cox proportional hazard models were used to estimate the hazard ratios (HRs) and 95% CIs for dementia mortality according to categories of olive oil intake, separately in each cohort. Participants were censored at death from causes other than dementia. Model 1 was stratified for age and calendar time. Multivariable model 2 was adjusted for Southern European/Mediterranean ancestry, married, living alone, smoking, alcohol intake, physical activity, multivitamin use, history of hypertension and hypercholesterolemia, in women postmenopausal status and menopausal hormone use, total energy intake, family history of dementia, history of depression, census SES, and BMI. Multivariable model 3 was further adjusted for intake of red meat, fruits and vegetables, nuts, soda, whole grains, and trans-fat, all indicative of diet quality.

In a secondary analysis we used the composite outcome for dementia-related deaths. We also repeated the main analysis in the genotyping subsample. We carried out mediation analyses to calculate the percentage of the association between olive oil intake and dementia-related mortality that is attributable to CVD, hypercholesterolemia, hypertension, and diabetes. We also performed stratified analyses by prespecified subgroups (eMethods in Supplement 1 ).

A joint analysis was performed to test whether olive oil intake (never or <1/mo, >0 to ≤7g/d, and >7g/d) and the AMED or the AHEI score (tertiles) combined as the exposure was associated with dementia mortality. In substitution analyses, we assessed the risk of dementia-related mortality by replacing 5 g/d of different fat sources, including margarine, mayonnaise, butter, and a combination of other vegetable oils (corn, safflower, soybean, and canola), with olive oil. Both continuous variables as 5-g/d increments were included in a multivariable model 3, mutually adjusted for other types of fat. The difference in the coefficients obtained for olive oil and the substituted fat provided the estimated HR and 95% CI for substituting 5 g/d of olive oil for an equivalent amount of the other fats.

Several exploratory sensitivity analyses were performed including a 4-year lagged analysis, analyses adjusting for other covariates, a cause-specific competing risk model and analyses excluding participants who self-reported having dementia at baseline (n = 12) (eMethods in Supplement 1 ). Analyses were performed from May 2022 to July 2023 using SAS version 9.4 (SAS Institute). All statistical tests were 2-sided with an α = .05.

Over 2 183 095 person-years of follow-up, this study documented a total of 4751 dementia deaths (3473 in NHS and 1278 in HPFS; 37 649 total deaths). Among 92 383 participants included at baseline in 1990, 60 582 (65.6%) were women, and the mean (SD) age was 56.4 (8.0) years. Mean (SD) olive oil intake was 1.3 (2.5) g/d in both NHS and HPFS; the mean (SD) adherence score for the Mediterranean diet was 4.5 (1.9) points in the NHS and 4.2 (1.9) points in the HPFS; and the mean (SD) AHEI diet quality score was 52.5 (11.1) points in the NHS and 53.4 (11.6) points in the HPFS.

Table 1 shows baseline characteristics of participants categorized by total olive oil intake. Participants with a higher olive oil intake (>7 g/d) at baseline had an overall higher caloric intake, but not a higher BMI, had better diet quality, had higher alcohol intake, were more physically active, and were less likely to smoke compared with those never consuming olive oil or less than once per month ( Table1 ). Individuals who were homozygous for the APOE ε4 allele were 5.5 to 9.4 times more likely to die with dementia compared with noncarriers for the APOE e4 allele (χ 2  P  < .001) (eTable 1 in Supplement 1 ).

Olive oil intake was inversely associated with dementia-related mortality in age-stratified and multivariable-adjusted models ( Table 2 ). Compared with participants with the lowest olive oil intake, the pooled HR for dementia-related death among participants with the highest olive oil intake (>7 g/d) was 0.72 (95% CI, 0.64-0.81), after adjusting for sociodemographic and lifestyle factors. The association between each 5-g increment in olive oil consumption with dementia-related death was also inverse and significant in the pooled analysis. The multivariable-adjusted HR for dementia-related death for the highest compared with the lowest olive oil intake (>7 g/d) was 0.67 (95% CI, 0.59-0.77) for women and 0.87 (95% CI, 0.69-1.09) for men ( Table 2 ). Olive oil intake in 5-g increments was inversely associated with dementia-related mortality in women (HR, 0.88 [95% CI, 0.84-0.93]), but not in men (HR, 0.96 [95% CI, 0.88-1.04]). Analyses remained consistent when using the composite outcome for death with dementia (eTable 2 in Supplement 2 ). In the genotyping subsample, the results remained unchanged after further adjusting for the APOE ε4 allelic genotype (multivariable-adjusted pooled HR comparing high vs low olive oil intake, 0.66 [95% CI, 0.54-0.81]; P for trend < .001) (eTable 4 in Supplement 1 ). Pooled mediation analyses found that CVD, hypercholesterolemia, hypertension, and diabetes did not significantly attenuate the association (unchanged HRs with and without adjusting for the intermediate; data not shown).

In joint analyses, participants with the highest olive oil intake had a lower risk for dementia-related mortality, irrespective of their AMED score (28% to 34% lower risk compared with participants in the combined low olive oil and high AMED) ( Figure 1 A; eTable 3 in Supplement 1 ) and of their AHEI (27% to 38% lower risk compared with participants with low olive oil and high AHEI) ( Figure 1 B; eTable 3 in Supplement 1 ).

Replacing 5 g/d of mayonnaise with 5 g/d of olive oil was associated with a 14% (95% CI, 7%-20%) lower risk of dementia-related mortality in pooled multivariable-adjusted models ( Figure 2 ). As for the substitution of 5 g/d of margarine with the equivalent amount of olive oil, we estimated an 8% (95% CI, 4%-12%) lower risk. Substitutions of other vegetable oils or butter with olive oil were not statistically significant.

Exploratory subgroup analyses (eFigure in Supplement 1 ) showed associations between higher olive oil intake and lower risk of dementia-related mortality across most subgroups. No statistically significant associations were found in participants with a family history of dementia, living alone, using a multivitamin, and in non– APOE ε4 carriers. Results from exploratory sensitivity analyses (eTables 5-8 in Supplement 1 ) were comparable with the findings from the main analysis (eResults in Supplement 1 ).

In 2 large US prospective cohorts of men and women, we found that participants who consumed more than 7 g/d of olive oil had 28% lower risk of dying from dementia compared with participants who never or rarely consumed olive oil. This association remained significant after adjustment for diet quality scores including adherence to the Mediterranean diet. We estimated that substituting 5 g/d of margarine and mayonnaise with olive oil was associated with significantly lower dementia-related death risk, but not when substituting butter and other vegetable oils. These findings provide evidence to support dietary recommendations advocating for the use of olive oil and other vegetable oils as a potential strategy to maintain overall health and prevent dementia.

In the NHS and HPFS, a lower risk of neurodegenerative disease mortality, including dementia mortality, was observed with higher olive oil consumption (HR, 0.81 [95% CI, 0.78-0.84]). 11 Evidence that pertains to cognitive decline or incident dementia is more widely available than it is for dementia mortality. 6 , 22 In the French Three-City Study (n = 6947), participants with the highest olive oil intake were 17% (95% CI, 1%-29%) less likely to experience a 4-year cognitive decline related to visual memory, but no association was found for verbal fluency (odds ratio [OR], 0.85 [95% CI, 0.70-1.03]). 22 Furthermore, participants with a higher intake of olive oil (moderate or intensive vs never) had a lower risk of verbal fluency and visual memory cognitive impairment. Potential sex differences were not investigated. In the PREDIMED trial, which supplemented a Mediterranean-style diet with extra-virgin olive oil (1 L/wk/household) or nuts (30 g/d), 23 the authors investigated cognitive effects and status in 285 and 522 cognitively healthy participants using global and in-depth neuropsychological battery testing. Although the study was not originally designed for cognitive outcomes and the effect of olive oil cannot be isolated, after 6.5 years, the olive oil group exhibited improved cognitive performance in verbal fluency and memory tests compared with a low-fat diet (control), and they were less prone to develop mild cognitive impairment (OR, 0.34 [95% CI, 0.12-0.97]; n = 285). 6 Global cognitive performance was higher in both the olive oil and the nut groups compared with the control post trial (n = 522). 8 These studies were conducted in Europe, in populations with typically higher olive oil intake compared with US populations.

Observational studies and some trials have consistently found associations between following diets such as the Mediterranean, DASH, MIND, and AHEI, and prudent patterns to healthier brain structure, 24 reduced cognitive impairment and Alzheimer risk, and improved cognitive function. 4 In our study, those with the highest olive oil intake (>7 g/d) had the lowest dementia-related death risk compared with those with minimal intake (never or less than once per month), regardless of diet quality. This highlights a potentially specific role for olive oil. Still, the group with both high AHEI scores and high olive oil intake exhibited the lowest dementia mortality risk (HR, 0.68 [95% CI, 0.58-0.79]; reference: low AHEI score and low olive oil intake), suggesting that combining higher diet quality with higher olive oil intake may confer enhanced benefit.

Olive oil consumption may lower dementia mortality by improving vascular health. 18 Several clinical trials support the effect of olive oil in reducing CVD via improved endothelial function, coagulation, lipid metabolism, oxidative stress, platelet aggregation and decreased inflammation. 25 Nonetheless, the results of our study remained independent of hypertension and hypercholesterolemia. Mild cognitive impairment, Alzheimer disease, and related dementias were associated with abnormal blood brain barrier permeability, possibly allowing the crossing of neurotoxic molecules into the brain. 26 Mechanistical evidence from animal 27 - 29 and human studies 9 , 30 have shown that phenolic compounds in olive oil, particularly extra-virgin olive oil, may attenuate inflammation, oxidative stress and restore blood brain barrier function, thereby reducing brain amyloid-β and tau-related pathologies and improving cognitive function. However, incident CVD, hypercholesterolemia, hypertension, and diabetes were not significant mediators of the association between olive oil intake and dementia-related death in our study.

The association was significant in both sexes but did not remain in men after full adjustment of the model. Some previous research has reported cognitive-related sex differences. Evidence from trials also showed sex- and/or gender-specific responses to lifestyle interventions for preventing cognitive decline, possibly due to differences in brain structure, hormones (sex) and social factors (gender). 31 Olive oil intake may be protective of dementia and related mortality, particularly in women. Nonetheless, we did not observe significant heterogeneity or interaction of cohort by olive oil intake on the risk of fatal dementia. Sex and gender differences should be carefully considered in future studies examining the association or effect of olive oil on cognitive-related outcomes to improve our understanding.

We found that using olive oil instead of margarine and mayonnaise, but not butter and other vegetable oils, was associated with a lower risk of dementia-related death. At the time of the study, margarine and mayonnaise contained considerable levels of hydrogenated trans-fats. The latter were strongly associated with all-cause mortality, CVD, type 2 diabetes, and dementia, 32 , 33 which may explain the lower dementia-related death risk observed when replacing it with olive oil. The US Food and Drug Administration banned manufacturers from adding partially hydrogenated oils to foods in 2020. 34 Future studies examining intake of trans-fat–free margarine will be informative. Although the substitution of butter with olive oil was found to be associated with a lower risk of type 2 diabetes, CVD, and total mortality, 11 we did not find an association with the risk of dementia mortality. Although these previous studies did not examine the associations for butter per se, intake of regular fat dairy products, including cheese, yogurt, and milk, was reported to be either not associated or inversely associated with lower cognitive function, cognitive decline, and dementia. 35 - 37

Our cohort analyses include several strengths, namely the long follow-up period and large sample size with a high number of dementia death cases. Also, we included genotyping of the APOE ε4 allele in a large subsample of participants to reduce potential confounding attributed to this well-known risk factor for Alzheimer disease. Additionally, our repeated diet measurements, weight, and lifestyle variables permitted us to account for long-term olive oil intake and confounding factors. Furthermore, the use of dietary cumulative average updates reduced random measurement error by considering within-person variations in intake.

This study has limitations. The possibility of reverse causation cannot be excluded due to the observational nature of our study. However, the 4-year lagged analysis results, consistent with the primary analysis, suggest that olive oil intake is predictive of dementia mortality rather than a consequence of premorbid dementia. While it is plausible that higher olive oil intake could be indicative of a healthier diet and higher SES, our results remained consistent after accounting for the latter. Despite adjusting for key covariates, residual confounding may remain due to unmeasured factors. Also, our study was conducted among health professionals. While this minimizes the potential confounding effects of socioeconomic factors and likely increases reporting due to a high level of education, this may also limit generalizability. Our population was predominantly of non-Hispanic White participants, limiting generalizability to more diverse populations. Additionally, we could not differentiate among various types of olive oil that differ in their polyphenols and other nonlipid bioactive compounds content.

This study found that in US adults, particularly women, consuming more olive oil was associated with lower risk of dementia-related mortality, regardless of diet quality. Substituting olive oil intake for margarine and mayonnaise was associated with lower risk of dementia mortality and may be a potential strategy to improve longevity free of dementia. These findings extend the current dietary recommendations of choosing olive oil and other vegetable oils to the context of cognitive health and related mortality.

Accepted for Publication: March 6, 2024.

Published: May 6, 2024. doi:10.1001/jamanetworkopen.2024.10021

Open Access: This is an open access article distributed under the terms of the CC-BY License . © 2024 Tessier AJ et al. JAMA Network Open .

Corresponding Authors: Anne-Julie Tessier, RD, PhD ( [email protected] ), and Marta Guasch-Ferré, PhD ( [email protected] ), Department of Nutrition, Harvard T.H. Chan School of Public Health, 655 Huntington Ave, Bldg 2, Boston, MA 02115.

Author Contributions: Drs Tessier and Guasch-Ferré had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Tessier, Chavarro, Hu, Willett, Guasch-Ferré.

Acquisition, analysis, or interpretation of data: Tessier, Cortese, Yuan, Bjornevik, Ascherio, Wang, Chavarro, Stampfer, Willett, Guasch-Ferré.

Drafting of the manuscript: Tessier.

Critical review of the manuscript for important intellectual content: Tessier, Cortese, Yuan, Bjornevik, Ascherio, Wang, Chavarro, Stampfer, Hu, Willett, Guasch-Ferré.

Statistical analysis: Tessier, Cortese, Wang, Willett, Guasch-Ferré.

Obtained funding: Chavarro, Stampfer, Hu, Guasch-Ferré.

Administrative, technical, or material support: Cortese, Yuan, Stampfer, Hu.

Supervision: Chavarro, Hu, Guasch-Ferré.

Conflict of Interest Disclosures: Dr Cortese reported a speaker honorarium from Roche outside the submitted work. Dr Ascherio reported receiving speaker honoraria from WebMD, Prada Foundation, Biogen, Moderna, Merck, Roche, and Glaxo-Smith-Kline. No other disclosures were reported.

Funding/Support: This study is supported by the research grant R21 AG070375 from the National Institutes of Health to Dr Guasch-Ferré. The NHS, NHSII and HPFS are supported by grants from the National Institutes of Health (UM1 CA186107, P01 CA87969, U01 CA167552, P30 DK046200, HL034594, HL088521, HL35464, HL60712). Dr Tessier is supported by the Canadian Institutes of Health Research (CIHR) Postdoctoral Fellowship Award. Dr Guasch-Ferré is supported the Novo Nordisk Foundation grant NNF23SA0084103.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data Sharing Statement: See Supplement 2 .

• Register for email alerts with links to free full-text articles
• Access PDFs of free articles
• Manage your interests
• Save searches and receive search alerts

• Share full article

Advertisement

Supported by

Study Suggests Genetics as a Cause, Not Just a Risk, for Some Alzheimer’s

People with two copies of the gene variant APOE4 are almost certain to get Alzheimer’s, say researchers, who proposed a framework under which such patients could be diagnosed years before symptoms.

By Pam Belluck

Scientists are proposing a new way of understanding the genetics of Alzheimer’s that would mean that up to a fifth of patients would be considered to have a genetically caused form of the disease.

Currently, the vast majority of Alzheimer’s cases do not have a clearly identified cause. The new designation, proposed in a study published Monday, could broaden the scope of efforts to develop treatments, including gene therapy, and affect the design of clinical trials.

It could also mean that hundreds of thousands of people in the United States alone could, if they chose, receive a diagnosis of Alzheimer’s before developing any symptoms of cognitive decline, although there currently are no treatments for people at that stage.

The new classification would make this type of Alzheimer’s one of the most common genetic disorders in the world, medical experts said.

“This reconceptualization that we’re proposing affects not a small minority of people,” said Dr. Juan Fortea, an author of the study and the director of the Sant Pau Memory Unit in Barcelona, Spain. “Sometimes we say that we don’t know the cause of Alzheimer’s disease,” but, he said, this would mean that about 15 to 20 percent of cases “can be tracked back to a cause, and the cause is in the genes.”

The idea involves a gene variant called APOE4. Scientists have long known that inheriting one copy of the variant increases the risk of developing Alzheimer’s, and that people with two copies, inherited from each parent, have vastly increased risk.

The new study , published in the journal Nature Medicine, analyzed data from over 500 people with two copies of APOE4, a significantly larger pool than in previous studies. The researchers found that almost all of those patients developed the biological pathology of Alzheimer’s, and the authors say that two copies of APOE4 should now be considered a cause of Alzheimer’s — not simply a risk factor.

The patients also developed Alzheimer’s pathology relatively young, the study found. By age 55, over 95 percent had biological markers associated with the disease. By 65, almost all had abnormal levels of a protein called amyloid that forms plaques in the brain, a hallmark of Alzheimer’s. And many started developing symptoms of cognitive decline at age 65, younger than most people without the APOE4 variant.

“The critical thing is that these individuals are often symptomatic 10 years earlier than other forms of Alzheimer’s disease,” said Dr. Reisa Sperling, a neurologist at Mass General Brigham in Boston and an author of the study.

She added, “By the time they are picked up and clinically diagnosed, because they’re often younger, they have more pathology.”

People with two copies, known as APOE4 homozygotes, make up 2 to 3 percent of the general population, but are an estimated 15 to 20 percent of people with Alzheimer’s dementia, experts said. People with one copy make up about 15 to 25 percent of the general population, and about 50 percent of Alzheimer’s dementia patients.

The most common variant is called APOE3, which seems to have a neutral effect on Alzheimer’s risk. About 75 percent of the general population has one copy of APOE3, and more than half of the general population has two copies.

Alzheimer’s experts not involved in the study said classifying the two-copy condition as genetically determined Alzheimer’s could have significant implications, including encouraging drug development beyond the field’s recent major focus on treatments that target and reduce amyloid.

Dr. Samuel Gandy, an Alzheimer’s researcher at Mount Sinai in New York, who was not involved in the study, said that patients with two copies of APOE4 faced much higher safety risks from anti-amyloid drugs.

When the Food and Drug Administration approved the anti-amyloid drug Leqembi last year, it required a black-box warning on the label saying that the medication can cause “serious and life-threatening events” such as swelling and bleeding in the brain, especially for people with two copies of APOE4. Some treatment centers decided not to offer Leqembi, an intravenous infusion, to such patients.

Dr. Gandy and other experts said that classifying these patients as having a distinct genetic form of Alzheimer’s would galvanize interest in developing drugs that are safe and effective for them and add urgency to current efforts to prevent cognitive decline in people who do not yet have symptoms.

“Rather than say we have nothing for you, let’s look for a trial,” Dr. Gandy said, adding that such patients should be included in trials at younger ages, given how early their pathology starts.

Besides trying to develop drugs, some researchers are exploring gene editing to transform APOE4 into a variant called APOE2, which appears to protect against Alzheimer’s. Another gene-therapy approach being studied involves injecting APOE2 into patients’ brains.

The new study had some limitations, including a lack of diversity that might make the findings less generalizable. Most patients in the study had European ancestry. While two copies of APOE4 also greatly increase Alzheimer’s risk in other ethnicities, the risk levels differ, said Dr. Michael Greicius, a neurologist at Stanford University School of Medicine who was not involved in the research.

“One important argument against their interpretation is that the risk of Alzheimer’s disease in APOE4 homozygotes varies substantially across different genetic ancestries,” said Dr. Greicius, who cowrote a study that found that white people with two copies of APOE4 had 13 times the risk of white people with two copies of APOE3, while Black people with two copies of APOE4 had 6.5 times the risk of Black people with two copies of APOE3.

“This has critical implications when counseling patients about their ancestry-informed genetic risk for Alzheimer’s disease,” he said, “and it also speaks to some yet-to-be-discovered genetics and biology that presumably drive this massive difference in risk.”

Under the current genetic understanding of Alzheimer’s, less than 2 percent of cases are considered genetically caused. Some of those patients inherited a mutation in one of three genes and can develop symptoms as early as their 30s or 40s. Others are people with Down syndrome, who have three copies of a chromosome containing a protein that often leads to what is called Down syndrome-associated Alzheimer’s disease .

Dr. Sperling said the genetic alterations in those cases are believed to fuel buildup of amyloid, while APOE4 is believed to interfere with clearing amyloid buildup.

Under the researchers’ proposal, having one copy of APOE4 would continue to be considered a risk factor, not enough to cause Alzheimer’s, Dr. Fortea said. It is unusual for diseases to follow that genetic pattern, called “semidominance,” with two copies of a variant causing the disease, but one copy only increasing risk, experts said.

The new recommendation will prompt questions about whether people should get tested to determine if they have the APOE4 variant.

Dr. Greicius said that until there were treatments for people with two copies of APOE4 or trials of therapies to prevent them from developing dementia, “My recommendation is if you don’t have symptoms, you should definitely not figure out your APOE status.”

He added, “It will only cause grief at this point.”

Finding ways to help these patients cannot come soon enough, Dr. Sperling said, adding, “These individuals are desperate, they’ve seen it in both of their parents often and really need therapies.”

Pam Belluck is a health and science reporter, covering a range of subjects, including reproductive health, long Covid, brain science, neurological disorders, mental health and genetics. More about Pam Belluck

The Fight Against Alzheimer’s Disease

Alzheimer’s is the most common form of dementia, but much remains unknown about this daunting disease..

How is Alzheimer’s diagnosed? What causes Alzheimer’s? We answered some common questions .

A study suggests that genetics can be a cause of Alzheimer’s , not just a risk, raising the prospect of diagnosis years before symptoms appear.

Determining whether someone has Alzheimer’s usually requires an extended diagnostic process . But new criteria could lead to a diagnosis on the basis of a simple blood test .

The F.D.A. has given full approval to the Alzheimer’s drug Leqembi. Here is what to know about i t.

Alzheimer’s can make communicating difficult. We asked experts for tips on how to talk to someone with the disease .

Dr Connor Richardson - Reserve, Resilience, and Protective Factors PIA Year in Review Recap Dementia Researcher Blogs

• Life Sciences

Dr Connor Richardson narrates his blog written for Dementia Researcher. Connor reflects on the latest advancements in dementia research, focusing on cognitive reserve and resilience. Connor revisits seminal papers and discussions from a recent webinar moderated by Harriet Demnitz-King of University College London, highlighting the complexities of defining and measuring cognitive reserve. The discussions explored various theories and mechanisms such as Brain Maintenance and Brain Reserve, with a particular emphasis on resilience in maintaining cognitive functions despite aging and disease. Key findings from the year include the role of astrocyte reactivity in Alzheimer's disease and the protective influence of educational attainment on dementia, showcasing both new insights and persistent challenges in the field. The review underscores ongoing debates and the introduction of novel approaches in studying cognitive decline, suggesting a promising direction for future research. Find the original text, and narration here on our website. https://www.dementiaresearcher.nihr.ac.uk/blog-reserve-resilience-protective-factors-pia-year-in-review-recap/ Don’t forget, you can get involved in the RRPF PIA by joining ISTAART and get access to previous webinars! -- Dr Connor Richardson is a Neuro-epidemiology Research Associate (soon to be NIHR Research Fellow) in the Newcastle University Population Health Sciences Institute. Connor is the research statistician for the Cognitive Function and Ageing studies (CFAS) multi-centre population cohort. His research interest lies in using advanced statistical modelling and machine learning to measure dementia risk. Connor blogs about his research, Equality, Diversity and Inclusion and sometimes his Pomapoo’s. -- Enjoy listening? We're always looking for new bloggers, drop us a line. http://www.dementiaresearcher.nihr.ac.uk This podcast is brought to you in association with Alzheimer's Association, Alzheimer's Research UK, Alzheimer's Society and Race Against Dementia, who we thank for their ongoing support. -- Follow us on Social Media: https://www.instagram.com/dementia_researcher/ https://www.facebook.com/Dementia.Researcher/ https://twitter.com/demrescommunity https://www.linkedin.com/company/dementia-researcher

• Episode Website
• More Episodes
• Copyright 2023 All rights reserved.

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Published: 06 May 2024

APOE4 homozygozity represents a distinct genetic form of Alzheimer’s disease

• Juan Fortea   ORCID: orcid.org/0000-0002-1340-638X 1 , 2 , 3   na1 ,
• Jordi Pegueroles   ORCID: orcid.org/0000-0002-3554-2446 1 , 2 ,
• Daniel Alcolea   ORCID: orcid.org/0000-0002-3819-3245 1 , 2 ,
• Olivia Belbin   ORCID: orcid.org/0000-0002-6109-6371 1 , 2 ,
• Oriol Dols-Icardo   ORCID: orcid.org/0000-0003-2656-8748 1 , 2 ,
• Lídia Vaqué-Alcázar 1 , 4 ,
• Laura Videla   ORCID: orcid.org/0000-0002-9748-8465 1 , 2 , 3 ,
• Juan Domingo Gispert 5 , 6 , 7 , 8 , 9 ,
• Marc Suárez-Calvet   ORCID: orcid.org/0000-0002-2993-569X 5 , 6 , 7 , 8 , 9 ,
• Sterling C. Johnson   ORCID: orcid.org/0000-0002-8501-545X 10 ,
• Reisa Sperling   ORCID: orcid.org/0000-0003-1535-6133 11 ,
• Alexandre Bejanin   ORCID: orcid.org/0000-0002-9958-0951 1 , 2 ,
• Alberto Lleó   ORCID: orcid.org/0000-0002-2568-5478 1 , 2 &
• Víctor Montal   ORCID: orcid.org/0000-0002-5714-9282 1 , 2 , 12   na1  

Nature Medicine ( 2024 ) Cite this article

11k Accesses

4199 Altmetric

Metrics details

• Alzheimer's disease
• Predictive markers

This study aimed to evaluate the impact of APOE4 homozygosity on Alzheimer’s disease (AD) by examining its clinical, pathological and biomarker changes to see whether APOE4 homozygotes constitute a distinct, genetically determined form of AD. Data from the National Alzheimer’s Coordinating Center and five large cohorts with AD biomarkers were analyzed. The analysis included 3,297 individuals for the pathological study and 10,039 for the clinical study. Findings revealed that almost all APOE4 homozygotes exhibited AD pathology and had significantly higher levels of AD biomarkers from age 55 compared to APOE3 homozygotes. By age 65, nearly all had abnormal amyloid levels in cerebrospinal fluid, and 75% had positive amyloid scans, with the prevalence of these markers increasing with age, indicating near-full penetrance of AD biology in APOE4 homozygotes. The age of symptom onset was earlier in APOE4 homozygotes at 65.1, with a narrower 95% prediction interval than APOE3 homozygotes. The predictability of symptom onset and the sequence of biomarker changes in APOE4 homozygotes mirrored those in autosomal dominant AD and Down syndrome. However, in the dementia stage, there were no differences in amyloid or tau positron emission tomography across haplotypes, despite earlier clinical and biomarker changes. The study concludes that APOE4 homozygotes represent a genetic form of AD, suggesting the need for individualized prevention strategies, clinical trials and treatments.

This is a preview of subscription content, access via your institution

Access options

Access Nature and 54 other Nature Portfolio journals

Get Nature+, our best-value online-access subscription

24,99 € / 30 days

cancel any time

Subscribe to this journal

Receive 12 print issues and online access

195,33 € per year

only 16,28 € per issue

Buy this article

• Purchase on Springer Link
• Instant access to full article PDF

Prices may be subject to local taxes which are calculated during checkout

Similar content being viewed by others

Exceptionally low likelihood of Alzheimer’s dementia in APOE2 homozygotes from a 5,000-person neuropathological study

Association of APOE e2 genotype with Alzheimer’s and non-Alzheimer’s neurodegenerative pathologies

Apolipoprotein E and Alzheimer disease: pathobiology and targeting strategies

Data availability.

Access to tabular data from ADNI ( https://adni.loni.usc.edu/ ), OASIS ( https://oasis-brains.org/ ), A4 ( https://ida.loni.usc.edu/collaboration/access/appLicense.jsp ) and NACC ( https://naccdata.org/ ) can be requested online, as publicly available databases. All requests will be reviewed by each studyʼs scientific board. Concrete inquiries to access the WRAP ( https://wrap.wisc.edu/data-requests-2/ ) and ALFA + ( https://www.barcelonabeta.org/en/alfa-study/about-the-alfa-study ) cohort data can be directed to each study team for concept approval and feasibility consultation. Requests will be reviewed to verify whether the request is subject to any intellectual property.

Code availability

All statistical analyses and raw figures were generated using R (v.4.2.2). We used the open-sourced R packages of ggplot2 (v.3.4.3), dplyr (v.1.1.3), ggstream (v.0.1.0), ggpubr (v.0.6), ggstatsplot (v.0.12), Rmisc (v.1.5.1), survival (v.3.5), survminer (v.0.4.9), gtsummary (v.1.7), epitools (v.0.5) and statsExpression (v.1.5.1). Rscripts to replicate our findings can be found at https://gitlab.com/vmontalb/apoe4-asdad (ref. 32 ). For neuroimaging analyses, we used Free Surfer (v.6.0) and ANTs (v.2.4.0).

Bellenguez, C. et al. New insights into the genetic etiology of Alzheimer’s disease and related dementias. Nat. Genet. 54 , 412–436 (2022).

Article   CAS   PubMed   PubMed Central   Google Scholar  

Frisoni, G. B. et al. The probabilistic model of Alzheimer disease: the amyloid hypothesis revised. Nat. Rev. Neurosci. 23 , 53–66 (2022).

Article   CAS   PubMed   Google Scholar  

Bateman R. J. et al. Clinical and biomarker changes in dominantly inherited Alzheimer’s disease. N. Engl. J. Med. 367 , 795–804 (2012).

Genin, E. et al. APOE and Alzheimer disease: a major gene with semidominant inheritance. Mol. Psychiatry 16 , 903–907 (2011).

Fortea, J. et al. Alzheimer’s disease associated with Down syndrome: a genetic form of dementia. Lancet Neurol. 20 , 930–942 (2021).

Fortea, J. et al. Clinical and biomarker changes of Alzheimer’s disease in adults with Down syndrome: a cross-sectional study. Lancet 395 , 1988–1997 (2020).

Jansen, W. J. et al. Prevalence of cerebral amyloid pathology in persons without dementia: a meta-analysis. JAMA 313 , 1924–1938 (2015).

Article   PubMed   PubMed Central   Google Scholar  

Saddiki H. et al. Age and the association between apolipoprotein E genotype and Alzheimer disease: a cerebrospinal fluid biomarker-based case-control study. PLoS Med. https://doi.org/10.1371/JOURNAL.PMED.1003289 (2020).

Jack, C. R. et al. NIA‐AA Research Framework: toward a biological definition of Alzheimer’s disease. Alzheimer’s Dement. 14 , 535–562 (2018).

Article   Google Scholar  

Beekly, D. L. et al. The National Alzheimer’s Coordinating Center (NACC) Database: an Alzheimer disease database. Alzheimer Dis. Assoc. Disord. 18 , 270–277 (2004).

PubMed   Google Scholar  

Montine, T. J. et al. National Institute on Aging–Alzheimer’s Association guidelines for the neuropathologic assessment of Alzheimer’s disease: a practical approach. Acta Neuropathol. 123 , 1–11 (2012).

Reiman, E. M. et al. Exceptionally low likelihood of Alzheimer’s dementia in APOE2 homozygotes from a 5,000-person neuropathological study. Nat. Commun. 11 , 1–11 (2020).

Iulita M. F. et al. Association of Alzheimer disease with life expectancy in people with Down syndrome. JAMA Netw. Open https://doi.org/10.1001/JAMANETWORKOPEN.2022.12910 (2022).

Corder, E. H. et al. Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer’s disease in late onset families. Science 261 , 921–923 (1993).

Fortea, J., Quiroz, Y. T. & Ryan, N. S. Lessons from Down syndrome and autosomal dominant Alzheimer’s disease. Lancet Neurol. 22 , 5–6 (2023).

Therriault, J. et al. Frequency of biologically defined Alzheimer’s disease in relation to age, sex, APOE ε4, and cognitive impairment. Neurology 96 , e975–e985 (2021).

Betthauser, T. J. et al. Multi-method investigation of factors influencing amyloid onset and impairment in three cohorts. Brain 145 , 4065–4079 (2022).

Snellman, A. et al. APOE ε4 gene dose effect on imaging and blood biomarkers of neuroinflammation and beta-amyloid in cognitively unimpaired elderly. Alzheimers Res. Ther. 15 , 71 (2023).

Ghisays, V. et al. Brain imaging measurements of fibrillar amyloid-β burden, paired helical filament tau burden, and atrophy in cognitively unimpaired persons with two, one, and no copies of the APOE ε4 allele. Alzheimers Dement. 16 , 598–609 (2020).

Mehta, R. I. & Schneider, J. A. What is ‘Alzheimer’s disease’? The neuropathological heterogeneity of clinically defined Alzheimer’s dementia. Curr. Opin. Neurol. 34 , 237–245 (2021).

van der Lee, S. J. et al. The effect of APOE and other common genetic variants on the onset of Alzheimer’s disease and dementia: a community-based cohort study. Lancet Neurol. 17 , 434–444 (2018).

Belloy, M. E., Napolioni, V. & Greicius, M. D. A quarter century of APOE and Alzheimera’s disease: progress to date and the path forward. Neuron 101 , 820–838 (2019).

Belloy, M. E. et al. APOE genotype and Alzheimer disease risk across age, sex, and population ancestry. JAMA Neurol. 80 , 1284–1294 (2023).

Jack, C. R. et al. Long-term associations between amyloid positron emission tomography, sex, apolipoprotein E and incident dementia and mortality among individuals without dementia: hazard ratios and absolute risk. Brain Commun. 4 , fcac017 (2022).

Morris, J. C. The Clinical Dementia Rating (CDR): current version and scoring rules. Neurology 43 , 2412–2414 (1993).

Weiner, M. W. et al. The Alzheimer’s Disease Neuroimaging Initiative 3: continued innovation for clinical trial improvement. Alzheimer’s Dement. 13 , 561–571 (2017).

Sperling R. A. et al. The A4 Study: stopping AD before symptoms begin? Sci. Transl. Med. https://doi.org/10.1126/scitranslmed.3007941 (2014).

Molinuevo, J. L. et al. The ALFA project: a research platform to identify early pathophysiological features of Alzheimer’s disease. Alzheimer’s Dement.: Transl. Res. Clin. Interventions 2 , 82–92 (2016).

Johnson, S. C. et al. The Wisconsin Registry for Alzheimer’s Prevention: a review of findings and current directions. Alzheimer’s Dement.: Diagnosis, Assess. Dis. Monit. 10 , 130–142 (2018).

Google Scholar  

LaMontagne P. J. et al. OASIS-3: longitudinal neuroimaging, clinical and cognitive dataset for normal aging and Alzheimer disease. Preprint at MedRxiv https://doi.org/10.1101/2019.12.13.19014902 (2019).

La Joie, R. et al. Multisite study of the relationships between antemortem [ 11 C]PIB-PET Centiloid values and postmortem measures of Alzheimer’s disease neuropathology. Alzheimers Dement. 15 , 205–216 (2019).

Montal, V. APOE4-ASDAD. GitLab https://gitlab.com/vmontalb/apoe4-asdad (2024).

Download references

Acknowledgements

We acknowledge the contributions of several consortia that provided data for this study. We extend our appreciation to the NACC, the Alzheimer’s Disease Neuroimaging Initiative, The A4 Study, the ALFA Study, the Wisconsin Register for Alzheimer’s Prevention and the OASIS3 Project. Without their dedication to advancing Alzheimer’s disease research and their commitment to data sharing, this study would not have been possible. We also thank all the participants and investigators involved in these consortia for their tireless efforts and invaluable contributions to the field. We also thank the institutions that funded this study, the Fondo de Investigaciones Sanitario, Carlos III Health Institute, the Centro de Investigación Biomédica en Red sobre Enfermedades Neurodegenerativas and the Generalitat de Catalunya and La Caixa Foundation, as well as the NIH, Horizon 2020 and the Alzheimer’s Association, which was crucial for this research. Funding: National Institute on Aging. This study was supported by the Fondo de Investigaciones Sanitario, Carlos III Health Institute (INT21/00073, PI20/01473 and PI23/01786 to J.F., CP20/00038, PI22/00307 to A.B., PI22/00456 to M.S.-C., PI18/00435 to D.A., PI20/01330 to A.L.) and the Centro de Investigación Biomédica en Red sobre Enfermedades Neurodegenerativas Program 1, partly jointly funded by Fondo Europeo de Desarrollo Regional, Unión Europea, Una Manera de Hacer Europa. This work was also supported by the National Institutes of Health grants (R01 AG056850; R21 AG056974, R01 AG061566, R01 AG081394 and R61AG066543 to J.F., S10 OD025245, P30 AG062715, U54 HD090256, UL1 TR002373, P01 AG036694 and P50 AG005134 to R.S.; R01 AG027161, R01 AG021155, R01 AG037639, R01 AG054059; P50 AG033514 and P30 AG062715 to S.J.) and ADNI (U01 AG024904), the Department de Salut de la Generalitat de Catalunya, Pla Estratègic de Recerca I Innovació en Salut (SLT006/17/00119 to J.F.; SLT002/16/00408 to A.L.) and the A4 Study (R01 AG063689, U24 AG057437 to R.A.S). It was also supported by Fundación Tatiana Pérez de Guzmán el Bueno (IIBSP-DOW-2020-151 o J.F.) and Horizon 2020–Research and Innovation Framework Programme from the European Union (H2020-SC1-BHC-2018-2020 to J.F.; 948677 and 847648 to M.S.-C.). La Caixa Foundation (LCF/PR/GN17/50300004 to M.S.-C.) and EIT Digital (Grant 2021 to J.D.G.) also supported this work. The Alzheimer Association also participated in the funding of this work (AARG-22-923680 to A.B.) and A4/LEARN Study AA15-338729 to R.A.S.). O.D.-I. receives funding from the Alzheimer’s Association (AARF-22-924456) and the Jerome Lejeune Foundation postdoctoral fellowship.

Author information

These authors contributed equally: Juan Fortea, Víctor Montal.

Authors and Affiliations

Sant Pau Memory Unit, Hospital de la Santa Creu i Sant Pau - Biomedical Research Institute Sant Pau, Barcelona, Spain

Juan Fortea, Jordi Pegueroles, Daniel Alcolea, Olivia Belbin, Oriol Dols-Icardo, Lídia Vaqué-Alcázar, Laura Videla, Alexandre Bejanin, Alberto Lleó & Víctor Montal

Centro de Investigación Biomédica en Red de Enfermedades Neurodegenerativas. CIBERNED, Barcelona, Spain

Juan Fortea, Jordi Pegueroles, Daniel Alcolea, Olivia Belbin, Oriol Dols-Icardo, Laura Videla, Alexandre Bejanin, Alberto Lleó & Víctor Montal

Barcelona Down Medical Center, Fundació Catalana Síndrome de Down, Barcelona, Spain

Juan Fortea & Laura Videla

Department of Medicine, Faculty of Medicine and Health Sciences, Institute of Neurosciences, University of Barcelona, Barcelona, Spain

Lídia Vaqué-Alcázar

Barcelonaβeta Brain Research Center (BBRC), Pasqual Maragall Foundation, Barcelona, Spain

Juan Domingo Gispert & Marc Suárez-Calvet

Neurosciences Programme, IMIM - Hospital del Mar Medical Research Institute, Barcelona, Spain

Department of Medicine and Life Sciences, Universitat Pompeu Fabra, Barcelona, Spain

Centro de Investigación Biomédica en Red Bioingeniería, Biomateriales y Nanomedicina. Instituto de Salud carlos III, Madrid, Spain

Centro Nacional de Investigaciones Cardiovasculares (CNIC), Madrid, Spain

Wisconsin Alzheimer’s Disease Research Center, University of Wisconsin-Madison School of Medicine and Public Health, Madison, WI, USA

Sterling C. Johnson

Brigham and Women’s Hospital Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA

Reisa Sperling

Barcelona Supercomputing Center, Barcelona, Spain

Víctor Montal

You can also search for this author in PubMed   Google Scholar

Contributions

J.F. and V.M. conceptualized the research project and drafted the initial manuscript. V.M., J.P. and J.F. conducted data analysis, interpreted statistical findings and created visual representations of the data. O.B. and O.D.-I. provided valuable insights into the genetics of APOE. L.V., A.B. and L.V.-A. meticulously reviewed and edited the manuscript for clarity, accuracy and coherence. J.D.G., M.S.-C., S.J. and R.S. played pivotal roles in data acquisition and securing funding. A.L. and D.A. contributed to the study design, offering guidance and feedback on statistical analyses, and provided critical review of the paper. All authors carefully reviewed the manuscript, offering pertinent feedback that enhanced the study’s quality, and ultimately approved the final version.

Corresponding authors

Correspondence to Juan Fortea or Víctor Montal .

Ethics declarations

Competing interests.

S.C.J. has served at scientific advisory boards for ALZPath, Enigma and Roche Diagnostics. M.S.-C. has given lectures in symposia sponsored by Almirall, Eli Lilly, Novo Nordisk, Roche Diagnostics and Roche Farma, received consultancy fees (paid to the institution) from Roche Diagnostics and served on advisory boards of Roche Diagnostics and Grifols. He was granted a project and is a site investigator of a clinical trial (funded to the institution) by Roche Diagnostics. In-kind support for research (to the institution) was received from ADx Neurosciences, Alamar Biosciences, Avid Radiopharmaceuticals, Eli Lilly, Fujirebio, Janssen Research & Development and Roche Diagnostics. J.D.G. has served as consultant for Roche Diagnostics, receives research funding from Hoffmann–La Roche, Roche Diagnostics and GE Healthcare, has given lectures in symposia sponsored by Biogen, Philips Nederlands, Esteve and Life Molecular Imaging and serves on an advisory board for Prothena Biosciences. R.S. has received personal consulting fees from Abbvie, AC Immune, Acumen, Alector, Bristol Myers Squibb, Janssen, Genentech, Ionis and Vaxxinity outside the submitted work. O.B. reported receiving personal fees from Adx NeuroSciences outside the submitted work. D.A. reported receiving personal fees for advisory board services and/or speaker honoraria from Fujirebio-Europe, Roche, Nutricia, Krka Farmacéutica and Esteve, outside the submitted work. A.L. has served as a consultant or on advisory boards for Almirall, Fujirebio-Europe, Grifols, Eisai, Lilly, Novartis, Roche, Biogen and Nutricia, outside the submitted work. J.F. reported receiving personal fees for service on the advisory boards, adjudication committees or speaker honoraria from AC Immune, Adamed, Alzheon, Biogen, Eisai, Esteve, Fujirebio, Ionis, Laboratorios Carnot, Life Molecular Imaging, Lilly, Lundbeck, Perha, Roche and outside the submitted work. O.B., D.A., A.L. and J.F. report holding a patent for markers of synaptopathy in neurodegenerative disease (licensed to Adx, EPI8382175.0). The remaining authors declare no competing interests.

Peer review

Peer review information.

Nature Medicine thanks Naoyuki Sato, Yadong Huang and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editor: Jerome Staal, in collaboration with the Nature Medicine team.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary information.

Supplementary Methods, Results, Bibliography, Figs. 1–7 and Tables 1–3.

Reporting Summary

Supplementary code.

This code is also available in the GitLab repository.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Cite this article.

Fortea, J., Pegueroles, J., Alcolea, D. et al. APOE4 homozygozity represents a distinct genetic form of Alzheimer’s disease. Nat Med (2024). https://doi.org/10.1038/s41591-024-02931-w

Download citation

Received : 03 November 2023

Accepted : 19 March 2024

Published : 06 May 2024

DOI : https://doi.org/10.1038/s41591-024-02931-w

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.