Alibaba open-sources medical AI model that can detect cancer and nearly 150 conditions | South China Morning PostAdvertisementAlibabaTechBig TechAlibaba open-sources medical AI model that can detect cancer and nearly 150 conditionsTested on nearly 40,000 real-world exams, the model outperformed most radiologists, according to a new study published in Science2-MIN READ2-MIN1 ListenAnn Caoin ShanghaiPublished: 10:30pm, 18 Sep 2026Alibaba Group Holding’s research arm, Damo Academy, has open-sourced an artificial intelligence model capable of identifying nearly 150 abdominal conditions – including cancers – by reading computed tomography (CT) scans, marking the latest step in the firm’s growing medical AI efforts.The vision-language model, called Damo Radar, was designed to analyse contrast-enhanced CT scans covering 18 abdominal organs and identify a broad range of diseases and other abnormalities, such as malignant tumours, the institute said on Friday.The model was trained using CT scans paired with clinical reports. In nearly 40,000 real-world examinations, it achieved an average area under the curve (AUC) of 0.913 across 146 clinical findings. An AUC of 1.0 represents perfect diagnostic accuracy.The research team said the training method could eventually be extended to other types of medical imaging, calling the model “the world’s first expert-level generalist medical imaging model”.Select VoiceSelect Speed0.8x0.9x1.0x1.1x1.2x1.5x1.75x00:0000:001xAI-generated voice

Alibaba Group Holding’s research arm, Damo Academy, has released an artificial intelligence model derived from their ongoing medical AI efforts, marking a significant step forward in diagnostic imaging. This model, named Damo Radar, is a vision-language model designed to analyze contrast-enhanced computed tomography (CT) scans encompassing eighteen abdominal organs to identify a broad spectrum of diseases and abnormalities, including nearly 150 abdominal conditions and various cancers. The model was developed by analyzing CT scans paired with corresponding clinical reports as the training data. In rigorous testing involving nearly 40,000 real-world examinations, the model demonstrated strong diagnostic capabilities, achieving an average area under the curve of 0.913 across 146 distinct clinical findings. This performance metric indicates that the model successfully processes complex medical imagery to detect a wide range of conditions with high accuracy. The research team posits that this methodology could be extended to other modalities of medical imaging, positioning the system as what they term the world’s first expert-level generalist medical imaging model.