China’s AI Doctor DeepRare Outperforms Experts in Detecting Rare Diseases

Beijing: A team of Chinese researchers has developed a new AI-powered tool named DeepRare that can rapidly and accurately diagnose rare and complex diseases, surpassing the performance of experienced doctors. According to a study published on February 18 in the journal Nature, DeepRare achieved 79% overall success in identifying diseases across 6,401 test cases, outperforming five senior doctors who correctly diagnosed only 66% of cases.

How DeepRare Works

DeepRare employs a large language model integrated into an agent-based framework. It leverages over 40 AI agents and tools to analyze medical notes, match symptoms to diseases, search relevant literature, and interpret genetic sequencing data. The tool generates a ranked list of potential diagnoses using verified medical references, reducing the need for manual literature reviews.

Surpassing Human Experts

In the study, DeepRare made correct first-attempt diagnoses in 64.4% of cases, compared to 54.6% for human experts. It also outperformed 15 existing diagnostic tools, including popular systems like Exomiser, by integrating patient data with genetic analysis for early detection.

Tackling Rare Diseases

Rare diseases affect fewer than 1 in 2,000 people, yet globally impact over 300 million individuals. Over 7,000 rare diseases exist, with approximately 80% having a genetic origin. DeepRare’s ability to quickly process complex data makes it a promising tool for early and accurate detection, potentially transforming patient care worldwide.

Implications for Healthcare

The success of DeepRare highlights the growing role of AI in medical diagnostics. Experts believe that tools like this could soon complement human doctors, speeding up detection, improving treatment outcomes, and saving lives, especially for conditions that are difficult to diagnose.

Bottom line: DeepRare demonstrates that AI is no longer just an academic tool—it is now a practical, high-performing assistant capable of outperforming seasoned medical professionals in detecting rare diseases.