Machine Learning-Based Prediction Model for Depression Tendency in Chinese Elderly

As global aging accelerates, the incidence of depression among the elderly has risen significantly, posing a serious threat not only to their quality of life but also to their overall physical health and well-being. While timely identification and effective intervention are crucial for addressing this issue, existing research often falls short in providing a systematic analysis of the various influencing factors. This study utilizes the China Health and Retirement Longitudinal Study (CHARLS) dataset to thoroughly explore the multiple factors that contribute to depression in older adults. During the data preprocessing phase, we meticulously ensured data quality by analyzing the relationships between variables through correlation coefficient heatmaps and conducting chi-square tests to assess independence. Following this, we constructed and optimized several predictive models, including random forests, decision trees, and naive Bayes, fine-tuning their parameters for enhanced performance. Ultimately, the experimental results demonstrated that the optimized Support Vector Machine (SVM) model excelled in predicting depression in the elderly, providing robust theoretical support and empirical evidence that can inform targeted strategies for improving the mental health and overall well-being of older adults.