Machine learning algorithms to predict depression in older adults in China: a cross-sectional study

Area of Science:

• Gerontology
• Computational Psychiatry
• Public Health

Background:

• Depression is a significant health concern for older adults globally.
• Understanding predictive factors for depression in China's aging population is crucial for targeted interventions.

Purpose of the Study:

• To evaluate the predictive capability of machine learning (ML) algorithms for depression incidence in China's older adults.
• To identify key factors contributing to depression within this demographic.

Main Methods:

• Utilized data from 7,880 older adults from the 2020 China Health and Retirement Longitudinal Study.
• Applied six ML algorithms (logistic regression, k-nearest neighbors, support vector machine, decision tree, LightGBM, random forest) for predictive modeling.
• Employed Delong test for ROC curve comparison and decision curve analysis (DCA) for model performance evaluation, with Shapely Additive exPlanations for interpretation.

Main Results:

• Models achieved an Area Under the Curve (AUC) range of 0.648-0.738, with significant differences observed (P < 0.01).
• LightGBM demonstrated the highest net benefit across various probability thresholds according to DCA.
• Top predictors for depression included self-rated health, nighttime sleep, gender, age, and cognitive function.

Conclusions:

• Machine learning algorithms effectively predict depression incidence in China's older adult population.
• Identified critical factors associated with depression, enabling targeted prevention and treatment strategies.