A Chinese research team from Xinhua Hospital and Shanghai Jiao Tong University has developed AI-based disease prediction system DeepRare, an AI tool that set a new global record for rare disease diagnostic accuracy by combining clinical data, genetic information, and evidence reasoning, according to a study published in Nature.
The AI diagnosed rare disease breakthrough addresses one of medicine’s most persistent challenges: diagnosing rare diseases quickly and accurately, especially in regions where genetic testing is limited.
Developed by researchers from Xinhua Hospital and the Shanghai Jiao Tong University School of Artificial Intelligence, AI-based disease prediction tool, DeepRare, combines clinical expertise with advanced AI for disease identification and diagnosis.
Last July, the AI-based disease prediction platform launched online, attracting more than 1,000 professional users from over 600 medical and research institutions worldwide, highlighting global interest in reliable diagnostic tools.
Test results show that when provided only with patients’ clinical phenotypic information without genetic data, DeepRare achieved a first-attempt accuracy rate of 57.18%, improving nearly 24 percentage points over the previous global model.
When genetic data were included, its diagnostic accuracy exceeded 70%, confirming its capability as AI-powered target identification for rare diseases.
AI In Medical Imaging Diagnosis Future Potential
Although DeepRare primarily addresses rare disease diagnosis, its evidence-based AI-based disease prediction reasoning model highlights the potential of AI in healthcare diagnostics. Traditional AI systems often face trust issues because their reasoning is difficult to trace.
DeepRare addresses this by providing every diagnostic result with a full chain of evidence, making it an AI smart diagnostic tool that doctors can rely on.
The AI-based disease prediction system’s strength lies in integrating real-time access to a vast database of medical literature and clinical case data. Its iterative cycle of hypothesis testing, verification, and self-reflection assesses diagnostic leads and fills logical gaps.
Researchers note that such transparency and reliability make it suitable for conditions suitable for AI-assisted diagnosis and expand its use in global medical practice.
Global Rare Disease Alliance
Rare disease diagnosis has long been hindered by limited data, fragmented expertise, and unequal access to genetic testing. DeepRare’s performance using only clinical data offers a practical solution for under-resourced regions, narrowing diagnostic disparities.
By improving outcomes without relying solely on lab infrastructure, the system exemplifies AI for disease identification and diagnosis at its most practical level.
Sun Kun, one of the corresponding authors, said the team is preparing to launch “a global AI alliance for rare disease diagnosis and treatment” and plans to validate 20,000 real-world cases within six months. Researchers aim to strengthen clinical confidence and advance AI-powered target identification for rare diseases, ultimately redefining how technology supports physicians and patients.
If successful, DeepRare may reshape rare disease care and illustrate how AI-powered target identification for rare diseases and AI diagnosed rare disease models can earn trust not as opaque decision-makers, but as transparent, accountable partners in patient care.
Inside Telecom provides you with an extensive list of content covering all aspects of the Tech industry. Keep an eye on our Medtech section to stay informed and updated with our daily articles.