Alibaba’s research arm, Damo Academy, open-sourced a new medical vision-language AI model named RADAR. The system analyzes contrast-enhanced abdominal CT scans across 18 organs to identify nearly 150 conditions, including various cancers.
The journal Science published the study detailing the model's performance. RADAR achieved high diagnostic accuracy in a test involving nearly 40,000 real-world exams.
The AI model’s average accuracy surpassed 23 out of 26 human radiologists in a direct comparison. As an assistive tool, RADAR reduced missed diagnoses by 10%. The technology also cut scan reading times by more than 30%.