Purpose/Significance To explore the characteristics and rules of doctor-patient dialogues in online health community, so as to improve the efficiency and effect of doctor-patient interaction in virtual space, and optimize the quality of knowledge service in online health community. Method/Process Doctor-patient interaction data from the “Haodaifu” website are obtained using web scraping technology, and the interaction patterns and behavioral characteristics of online health community are analyzed by adopting a combination of open coding and conversation analysis. Result/Conclusion 3 distinct interaction modes are identified according to the findings:traditional paternalistic, simple question-and-answer, and interactive consolation. The detailed description of initial symptoms provided by users, the proficiency in question-and-answer skills exhibited by doctors and patients, and the knowledge disparities between them are identified as the primary factors influencing interaction in online health community.