Predicting the long-term cognitive trajectories using machine learning approaches: A Chinese nationwide longitudinal...

Area of Science:

• Gerontology
• Cognitive Neuroscience
• Epidemiology

Background:

• Cognitive decline in aging populations is a significant public health concern.
• Understanding long-term cognitive trajectories and their determinants is crucial for targeted interventions.

Purpose of the Study:

• To explore long-term cognitive trajectories and their determinants in older Chinese adults.
• To construct predictive models for identifying individuals at high risk of unfavorable cognitive trajectories.

Main Methods:

• A 16-year longitudinal cohort study of 3502 older adults (aged 65-105).
• Cognitive function assessed using the Chinese version of the Mini Mental State Examination (MMSE).
• Machine learning algorithms (r-LR, SVM, RF, SL) employed for predictive modeling.

Main Results:

• Two distinct cognitive trajectories were identified: intact cognitive functioning (93.6%) and dementia (6.4%).
• Factors associated with rapid cognitive decline included older age, female gender, Han ethnicity, no schooling, rural residence, low leisure activity, and low baseline BADL score.
• Regularized logistic regression, SVM, and Super Learner demonstrated good predictive performance (Sensitivity: 0.73, G-mean: 0.65), with age and psychological well-being as key predictors.

Conclusions:

• Two distinct cognitive trajectories exist among older Chinese adults.
• Identified risk factors can inform the development of early risk prediction models for cognitive decline.