Purpose/Significance To build a knowledge graph for the rehabilitation of elderly diseases, in order to explore a home-based elderly self-management system that integrates prevention and rehabilitation, improve the personalized level and efficiency of the intelligent elderly disease rehabilitation and care system, and provide help for coping with the aging population. Method/Process Based on natural language processing (NLP) algorithms and the manual annotation from medical experts, a knowledge graph for geriatrics is built, which is performed by Med-BERT + BiLSTM + CRF combination algorithm, and an intelligent question answering system is established based on the knowledge graph.Result/Conclusion The quality of the newly designed knowledge graph for the rehabilitation of senile diseases meets the expectation, and the question answering system based on knowledge graph provides an interactive and friendly way for the elderly users to acquire knowledge.