Machine-assisted clinical decision support technology refers to the use of clinical medical records and medical knowledge to predict patients' diseases, assist doctors in disease diagnosis, and provide clinical decision support. Traditional expert rules to assist doctors in disease diagnosis and prediction have been put into practice, but the scheme has problems such as high update cost, low accuracy and weak robustness, which need to be overcome. iFLYHealth proposes a human-like learning and reasoning framework to achieve joint decision-making reasoning technology based on semantic evidence of medical record and semantics of knowledge graph. So far, we have achieved full coverage of common diseases at the community level, continuous accumulation of high-quality medical record libraries, and made remarkable achievements in community-level diagnosis and treatment.