머신러닝을 이용한 고령자의 우울군 판별모델 생성 및 우울 관련 요인 연구

초록

This study utilized machine learning to create a model for identifying elderly individuals at risk of depression and to assess the importance of variables affecting depression. The research data were obtained from the 2017 and 2020 Elderly Survey conducted by the Korea Institute for Health and Social Affairs, with a total of 19, 171 participants. The variables used in the study included personal characteristics, health status, health behaviors, and social support characteristics. Models were developed using Logistic Regression, Multi-Layer Perceptron, Support Vector Machine, Decision Tree, Random Forest, and XGBoost algorithms. The results indicated that the XGBoost model was the most optimal. The probability of being in the depression group increased with worse subjective health perception, poorer vision and chewing ability, poorer relationships with children, spouses, and friends, and higher IADL scores. Since subjective health status had the most significant impact on depression, it is recommended to implement mental health prevention programs targeting elderly individuals with poor subjective health perception. Mental health organizations can use this study to identify elderly individuals at higher risk of depression and provide targeted prevention and management, potentially reducing the incidence of depression and contributing to suicide prevention.

키워드