Abstract:In recent years, artificial intelligence has become a key direction of medical and health-related research and a hot spot of international competition. In order to investigate the current situation and challenges in hospital-led artificial intelligence researched, this study selects 14 national pilot hospitals to promote the high-quality development of public hospitals as samples, adopts a combination of quantitative and qualitative methods, analyzes the research articles related to artificial intelligence published by the sample hospitals in recent years, and analyzes the technical challenges in the hospital-led artificial intelligence research. The results show that although the number of hospital-led artificial intelligence research papers is increasing, in which 55% of the research is of prospect and expectation, while the quality of research could be improved. Meanwhile, the number of authorized AI related patent is rather small. The technical difficulties of hospital-led artificial intelligence research lie in the steep learning curve of artificial intelligence technology, high costs from computation iteration, difficulties in transferring clinical multimodal data into research data, and weak explainability. Hospitals should actively respond to policy promotion, reallocating resources to cultivate artificial intelligence coordinators, organize multi-modal data resources, and promote research and outputs.