Researchers in the British Journal of Cancer described a deep-learning-derived risk score intended to predict outcomes for patients treated with nivolumab in gastric cancer. Using pathology-derived inputs, the model aims to forecast how individuals may respond before immunotherapy begins. If validated prospectively, tumor-level image analytics could reduce trial-and-error treatment selection by identifying subsets likely to benefit from PD-1 blockade. For biotech, it reinforces the growing role of AI risk scoring in stratifying patients around expensive immunotherapies.