Using machine learning algorithms for predicting cognitive impairment and identifying modifiable factors among...

Area of Science:

• Gerontology
• Cognitive Science
• Machine Learning

Background:

• Cognitive impairment poses a significant challenge for aging populations worldwide.
• Early detection and identification of modifiable risk factors are crucial for intervention.

Purpose of the Study:

• To predict early-stage cognitive impairment in elderly Chinese individuals using machine learning.
• To identify reversible lifestyle factors associated with a slower rate of cognitive decline over three years.

Main Methods:

• Utilized data from 12,280 participants in the Chinese Longitudinal Healthy Longevity Survey (CLHLS) from 2002-2014.
• Employed six machine learning algorithms and an ensemble method to predict cognitive function based on the Mini-Mental State Examination (MMSE).
• Logistic regression analyzed the association between lifestyle behaviors and cognitive impairment risk.

Main Results:

• Support Vector Machine (SVM) and Multi-layer Perceptron (MLP) demonstrated high predictive performance (AUC 0.8267 and 0.8256).
• An ensemble model further improved predictive accuracy (AUC 0.8269).
• Increased engagement in activities such as playing Mahjong/cards, gardening, and watching TV/listening to the radio were associated with a reduced risk of cognitive impairment.

Conclusions:

• Machine learning, particularly SVM and ensemble models, effectively identifies elderly individuals at risk of cognitive impairment.
• Promoting leisure activities, gardening, and overall engagement may help mitigate cognitive decline in the elderly.