Dynamic Prediction Model Based on Relational Heterogeneous Graph Neural Networks

The data was mainly obtained from the website of China Health and Retirement Longitudinal Study. Regarding the clinical and behavioral habit factors related to sarcopenia, after cleaning and integration, the final clean data was obtained. Using these data, a dynamic prediction model for sarcopenia based on graph neural networks was constructed to achieve early warning of sarcopenia.