AI-enabled precision prediction and proactive management of cutaneous toxicities in cancer immunoradiotherapy (ICI+RT).

Immunoradiotherapy has become an increasingly important strategy for the treatment of advanced malignant tumors, but its broader application is accompanied by a high incidence of cutaneous toxicities, including radiation dermatitis and immune-related skin adverse events. These toxicities often emerge early during treatment, exhibit substantial inter-patient heterogeneity, and can compromise treatment continuity and patient quality of life. Conventional management remains largely reactive and grading-based, offering limited capacity for individualized risk assessment or early intervention. Recent advances highlight that cutaneous toxicity in immunoradiotherapy arises from the convergence of radiation-induced tissue injury and immune checkpoint blockade-driven immune amplification, involving interconnected pathways such as DNA damage-associated danger signaling, innate immune activation, cytokine amplification, and dysregulated T-cell effector responses. This biological complexity limits the predictive utility of single-factor or mechanism-isolated approaches, underscoring the need for integrative, data-driven strategies. In this mini-review, we synthesize emerging AI-enabled approaches for precision prediction and management of cutaneous toxicities in immunoradiotherapy. We focus on how clinicodosimetric variables, spatial dose topology, imaging- and radiomics-derived tissue susceptibility, and accessible immune-inflammatory surrogates can be integrated into pathway-informed predictive models. We further discuss translational frameworks that embed prediction into clinical workflows, enabling plan-aware exposure mitigation, proactive supportive care stratification, and dynamic on-treatment risk updating. Collectively, these advances position cutaneous toxicity as a tractable and clinically meaningful endpoint for precision management in immunoradiotherapy, aligned with the goals of data-driven oncology.