Abstract Artificial intelligence (AI) offers a promising solution to the long-standing challenge of accurately predicting treatment response in rectal cancer. In this narrative review, we summarize current AI-driven approaches to predicting pathologic complete response in rectal cancer. We also outline key barriers to clinical translation, including lack of standardization, small and geographically skewed training cohorts, domain shift across scanners and institutions, and broader ethical, regulatory, and medicolegal concerns. Finally, we highlight future directions, including federated learning to enable privacy-preserving multicenter model training, and emerging concepts such as virtual and digital twins that may support real-time adaptive therapy. These advances suggest that AI-based prediction of response in rectal cancer could be extremely valuable, but will require methodologically rigorous, multi-institutional efforts to be safely and equitably implemented.
Lopez et al. (Wed,) studied this question.