Machine learning models for predicting multimorbidity trajectories in middle-aged and elderly adults.

Area of Science:

• Gerontology
• Public Health
• Computational Biology

Background:

• Multimorbidity is a growing public health concern, particularly with global population aging.
• Effective prediction and management of multimorbidity progression in the elderly are essential.

Purpose of the Study:

• To develop predictive models for multimorbidity trajectories in middle-aged and elderly individuals.
• To identify key factors influencing multimorbidity progression.

Main Methods:

• Utilized data from the China Health and Retirement Longitudinal Study (CHARLS) database (12,198 participants aged 45+).
• Employed time-series clustering to define multimorbidity trajectories.
• Developed and evaluated machine learning models (XGBoost, Random Forest, SVM, Logistic Regression, ANN) for prediction.

Main Results:

• Identified four distinct multimorbidity progression patterns: Stable Low-Risk (45.26%), Progressively Worsening (14.35%), Moderate Stability (31.90%), and Consistently Deteriorating (8.49%).
• The XGBoost model demonstrated superior performance (Accuracy: 0.664, Macro ROC-AUC: 0.825).
• Key predictors identified include baseline disease counts, self-rated Activities of Daily Living (ADL), and self-rated health status.

Conclusions:

• Machine learning models can effectively predict multimorbidity trajectories in aging populations.
• Baseline disease count, ADL, and self-rated health are critical factors for predicting multimorbidity progression.