Can AI generate personalized chemotherapy regimens by analyzing tumor microenvironment images ?

Cancer treatment effectiveness depends on complex interactions between tumors and their surrounding tissues. AI can process high-resolution images of tumor microenvironments to identify therapeutic targets. Machine learning models could predict which chemotherapy drugs would be most effective for individual patients. This approach aims to move beyond one-size-fits-all treatment protocols. Clinical trials would be needed to validate these AI-generated regimens.

Today’s AI excels at detecting patterns in high-resolution histopathology images but does not autonomously design chemotherapy regimens; instead, it supports oncologists by predicting tumor subtypes, immune infiltration levels, or therapy response from microenvironment images. Cutting-edge pipelines combine deep-learning segmentation with multiparametric data (e.g., spatial transcriptomics) to score features like PD-L1 density or TLS maturity, which can be entered into clinical decision-support tools to suggest matching immunotherapies or combinations. However, AI outputs remain probabilistic and require prospective clinical trials before being used to choose cytotoxic drugs or dosing schedules. Regulatory frameworks for such “AI-informed prescribing” are still evolving.

— Enriched May 12, 2026 · Source: National Academies of Sciences, Engineering, and Medicine