Alibaba Launches RADAR: Outperforms 23 Out of 26 Radiologists and Detects 146 Diseases

Alibaba's DAMO Academy published its RADAR model in the peer-reviewed journal Science, after training it on more than 420,000 abdominal CT scans and demonstrating its ability to detect 146 diseases in a single scan, representing a milestone in clinical-scale medical AI application.

According to clinical evaluation, the model outperformed 23 out of 26 radiologists who participated in the study, an unprecedented figure in AI medical imaging. Publishing in Science sets a new benchmark for credibility in this field, as it is one of the world's premier peer-reviewed scientific journals.

Alibaba made the model fully open-source via GitHub without access restrictions or licensing fees, removing a key barrier to its adoption in resource-limited healthcare systems. The Arab region and Sub-Saharan Africa suffer from a severe shortage of radiologists, making this model a direct candidate for immediate implementation in public hospitals.

This release puts growing pressure on major medical imaging companies such as Siemens Healthineers and GE HealthCare, as well as medical AI regulatory bodies in the European Union and the U.S. Food and Drug Administration. For Arab health institutions, it opens the option to adopt this model directly in radiology departments without licensing costs, although implementation requires local verification of the model's accuracy across different demographic data.

What do these terms mean?

RADAR: An AI model developed by Alibaba's DAMO Academy, trained to analyze abdominal CT scans and detect 146 diseases in a single examination.

Open Source: Means that the model's code is freely available to everyone via GitHub, allowing any hospital or researcher to download and run it without paying licensing fees.

CT Scan: A medical imaging procedure that uses X-rays to capture detailed images of internal organs, used to detect diseases in the digestive system and abdomen.