Skip to main content

Table 4 Features selected from the MAS dataset by the best models using data-driven thresholding

From: Ensemble feature selection with data-driven thresholding for Alzheimer's disease biomarker discovery

MAS feature descriptions

AD biomarker evidence

Participant age at time of testing

The most important predictor of dementia [48]

Status of the epsilon4 allele of the APOE gene

Increases the risk of late-onset AD [11]

Waist to hip ratio

Obesity and cardiovascular risk factors have been linked to dementia [49]

Framingham cardio-vascular risk score

General Practitioner Assessment of Cognition score (GPCOG)

Designed to identify dementia [50]

Mini Mental State Exam score (MMSE)

Designed to identify dementia [51]

Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE)

Designed to identify dementia [52]

Informant subjective cognitive complaints – total score

Increasingly being recognised as predictors of progression to mild cognitive impairment and dementia [53]

Participant subjective cognitive complaints – total score

Composite variable encoding the number of major and minor at fault motor vehicle accidents in the past 18 months

Evidence exists that atypical changes in driving behaviours may be early signs of mild cognitive impairment (MCI) and dementia [54] [55]

Normal or abnormal posture

Abnormal posture or gait can be related to increasing frailty, which is associated with dementia [56]

Normal or abnormal gait

Urinary tract infection

Urinary tract infections are known to exacerbate dementia symptoms

Arthritis

A recent review highlighted the link between rheumatoid arthritis and dementia [57] although other researchers have reported a negative link with AD rather than dementia generally [58]

Urate

Studies have found a link between low serum uric acid levels and AD [59]

Uric acid