China’s new DeepRare AI system for rare disease diagnosis achieves record accuracy and offers evidence-based, traceable clinical reasoning.

SHANGHAI: A Chinese research team has developed DeepRare, an advanced AI system designed to improve rare disease diagnosis, with findings published in the journal Nature. The DeepRare AI system marks a significant step forward in clinical decision support, particularly in regions where access to genetic testing remains limited.

Also Read: China unveils high‑power microwave weapon that could target satellites

Developed by researchers from Xinhua Hospital, affiliated with Shanghai Jiao Tong University School of Medicine and its School of Artificial Intelligence, the platform aims to address long-standing diagnostic challenges. Rare diseases are often difficult to confirm without genetic data, and traditional AI tools have faced criticism for opaque reasoning processes that limit clinical trust.

Test results show that when only patients’ clinical phenotypic information was used, the DeepRare AI system achieved a first-attempt diagnostic accuracy of 57.18 per cent. This represents an improvement of nearly 24 percentage points over the previous leading global model. When genetic data were added, diagnostic accuracy exceeded 70 per cent.

Also Read: China’s light industry sees stable growth with trade-in boost and export gains

According to the research team, the system integrates real-time access to extensive medical literature and real-world clinical case data. It operates through an iterative reasoning cycle involving hypothesis generation, verification and self-reflection, allowing it to reassess evidence and correct logical gaps.

Each diagnosis generated by the DeepRare AI system includes a full chain of supporting evidence. This enables doctors to understand not only the conclusion but also the clinical reasoning behind it, improving transparency and confidence.

The online platform, launched last July, has already registered more than 1,000 professional users across over 600 medical and research institutions worldwide. Researchers are now preparing to establish a global AI alliance and aim to validate 20,000 rare disease cases within the next six months.