Insights into artificial intelligence-based digital pathology in hepatobiliary and pancreatic cancer

Artificial intelligence (AI)-driven digital pathology has emerged as a promising approach for improving diagnosis, prognostic assessment, and treatment decision-making in hepatobiliary and pancreatic cancers (HPCs). By leveraging whole-slide imaging and deep learning, AI enables high-throughput and reproducible analysis of complex histopathological data, facilitating precision oncology. This review summarizes recent advances in AI-assisted digital pathology across hepatocellular carcinoma, biliary tract cancer, and pancreatic cancer, with a focus on diagnostic applications, molecular prediction, prognostic modeling, and treatment response assessment. Despite rapid technical progress, the clinical translation of AI remains limited. Current challenges extend beyond model performance and include data heterogeneity, limited interpretability, lack of prospective validation, and insufficient integration into clinical workflows. Notably, the central limitation lies in the misalignment between algorithm development and clinically relevant problem definition. Future research should prioritize multicenter validation, multimodal data integration, and the development of biologically informed and clinically interpretable models. Importantly, AI is expected to evolve from a diagnostic support tool to a dynamic, workflow-integrated decision-support system. Overall, the impact of AI-driven digital pathology will depend not only on technological advancement, but on its ability to deliver measurable clinical benefit and to be effectively embedded into real-world clinical practice.