Abstract: 【Purpose/significance】Healthcare big data provides unprecedented opportunities for smart healthcare. However, "data
chimneys", "information silos" and inefficient knowledge service methods seriously hinder the innovation of medical and health ser‐
vice models. How to realize knowledge management innovation for all-round and full-cycle smart medical services through deep inte‐
gration of medical and health big data and dynamic knowledge services has become an important issue in the field of medical informa‐
tion resource management【. Method/process】We introduce a federated learning mechanism and deep fusion method for the secure shar‐ing of large-scale multi-source heterogeneous medical and health data, and propose a human-machine collaborative medical case da‐tabase construction method and a medical case knowledge recommendation method based on Jaccard distance algorithm【. Result/con‐clusion】This method provides an integrated knowledge management service framework for smart diagnosis and treatment, clinical teaching and auxiliary scientific research【. Innovation/limitation】This method not only provides a data management method system for smart medical and precise health management, but also provides a new idea for the construction of a new model of 5P smart medical service.