A new AI model called COMPASS is being positioned as a tool to improve predictions of patient response to immune checkpoint inhibitors. Developed by researchers at Harvard Medical School led by Associate Professor Marinka Zitnik, the work describes how the system uses data-driven signals to estimate which patients are more likely to benefit from specific immunotherapy drugs. COMPASS is built to tackle a recurring clinical bottleneck in oncology: checkpoint inhibitors can produce durable benefit in subsets of patients, but reliable pre-treatment predictors remain imperfect. Better stratification could influence trial design and treatment selection, especially as immunotherapy expands across tumor types. For the biotech industry, the immediate relevance is pipeline differentiation—AI-driven companion analytics can be deployed alongside therapeutic programs to sharpen target populations and potentially reduce trial failures. Key industry takeaway: COMPASS aims to bring higher-precision decision support to immune checkpoint therapy selection.