AI-driven pathology in esophageal cancer: from early screening to precision prognostics

Esophageal cancer (EC) represents a globally prevalent and highly aggressive malignancy, where early screening and precise diagnosis constitute foundational elements of contemporary therapeutic strategies, with pathologic review serving as the gold standard for definitive diagnosis and clinical assessment. This review comprehensively synthesizes recent advances in artificial intelligence (AI) models-leveraging diverse algorithmic frameworks-to augment pathological workflows across critical domains: early screening of Barrett's esophagus (BE) and incipient EC; diagnostic refinement through invasion depth quantification, histopathological subtyping, and molecular pathology analysis; metastatic evaluation of lymph node involvement; prognostic prediction of patient survival; and efficacy assessment for multimodal therapies. These AI-driven methodologies demonstrate significant clinical utility throughout the EC disease continuum. We further critically examine persistent performance challenges and societal implications, offering insights to inspire future research toward precision oncology and optimized pathological efficiency.