A Chinese research team recently developed an artificial intelligence (AI) system capable of diagnosing rare diseases, with related results published in the international journal Nature on the 19th. This system, named DeepRare, is hailed as the world's first rare disease AI doctor with a 'traceable reasoning process.'
For a long time, rare diseases have been difficult to diagnose. Previous medical AI systems could offer diagnoses, but lacked supporting evidence. This new system stores top-tier global medical knowledge and real case data. It first proposes a hypothesis, then seeks evidence for validation, and finally draws a conclusion. Test data shows that, even relying solely on the patient's clinical symptoms, the system attained a first-time diagnostic accuracy rate of 57.18%, which is nearly 24% higher than the previous international best model. If genetic data is included, the system's diagnostic accuracy rate can exceed 70%.
This system has operated an online diagnostic platform since July last year, and to date, over 600 medical institutions worldwide have registered to use it, ranging from major hospitals in China to top laboratories in Europe and the United States. Shanghai Xinhua Hospital will also officially launch this system to help doctors provide more comprehensive care for rare disease patients. Professor Sun Kun, one of the system's developers and affiliated with Shanghai Jiao Tong University School of Medicine's Xinhua Hospital, stated that the next step is to initiate a global rare disease diagnosis and treatment alliance. Over the next six months, the system will be further validated using 20,000 real cases.