In the medical human-computer interaction scenario, how to automatically interact with patients through AI to collect information so as to accurately answer patients' questions, and give reasonable suggestions and treatment options is an urgent problem for AI in the medical field. iFLYHealth innovatively combines the event knowledge graph and dynamic information of diagnosis and treatment together to better describe the rich logical relations (such as continuation, causality, conditional relation, hyponymy relation, etc.) among medical events and entities in spatial-temporal domain, correlates the events perceived during the conversation to the ones in the Event-Evolutionary Graph, then evolves and reasons, and finally, gives reasonable suggestions to the patients.