GMAsia
    🇨🇳China·AI News·18 Sept 2026·via SCMP

    Alibaba open-sources medical AI model that can detect cancer and nearly 150 conditions

    Alibaba's Damo Academy has open-sourced an AI model named Damo Radar. This vision-language model detects nearly 150 abdominal conditions, including cancers, from CT scans. It was trained on CT scans paired with clinical reports. The model achieved an average AUC of 0.913 across 146 clinical findings in nearly 40,000 real-world examinations, outperforming most radiologists. This marks a significant step in Alibaba's medical AI initiatives.

    Nexa's Summary

    Alibaba's Damo Radar, an open-sourced medical AI model, achieved an average AUC of 0.913 across 146 clinical findings. This performance, tested on nearly 40,000 real-world exams, suggests the model can surpass most human radiologists. The crucial detail is the open-sourcing: this democratizes access to advanced diagnostic capabilities, potentially accelerating medical AI development beyond Alibaba's immediate ecosystem.

    For China's healthcare system, Damo Radar offers a path to scale expert-level diagnostics. The model's ability to identify 150 conditions from CT scans could alleviate pressure on specialist radiologists, particularly in rural or underserved areas. This could enable more consistent and earlier detection of diseases like cancer, improving patient outcomes across the country. The open-source nature also invites collaboration from other Chinese tech firms and medical institutions.

    The thing to watch is how quickly Damo Radar's training method extends to other imaging types. The research team called it the "world's first expert-level generalist medical imaging model." If this generalization proves robust, it could set a new standard for AI diagnostics globally, not just in Asia. A key test will be its adoption and validation by independent medical bodies outside of Alibaba's direct influence.

    Original reporting by SCMPWe don't republish, read the full story â†’

    Related reading

    6 stories