A multimodal murmuration for immunotherapy

Integrating numerous multimodal, patient-level biomarkers — spanning genomic, transcriptomic, radiologic, and clinical data — significantly improves the prediction of cancer immunotherapy response compared with single-marker approaches. The study, published in Nature Medicine, demonstrates that combined models capture complementary biological signals missed by individual predictors. However, the authors emphasize that generalizability across diverse patient populations and tumor types remains a major unresolved challenge, limiting near-term clinical deployment.