Yonsei University Develops AI Model for Tumor Prediction

Researchers from Yonsei University have developed an innovative AI model named MSI-SEER, which accurately predicts microsatellite instability (MSI) and a tumor's responsiveness to immune checkpoint inhibitors (ICIs). This development was announced in a press release and is expected to significantly improve clinical outcomes for patients with gastric and colorectal cancers.

MSI-SEER utilizes a deep Gaussian process-based Bayesian model to analyze hematoxylin and eosin-stained whole-slide images. This approach allows for the prediction of MSI status in gastric and colorectal cancers, integrating uncertainty prediction to achieve state-of-the-art performance. The model also provides insights into ICI responsiveness by considering the stroma-to-tumor ratio.

The research team, including Jae-Ho Cheong and Jeonghyun Kang from Yonsei University College of Medicine, validated the model using large datasets from diverse racial backgrounds. The findings, published in the journal npj Digital Medicine, highlight the potential of MSI-SEER for real-world applications in precision cancer medicine.