A systematic review and meta-analysis evaluated radiomics and artificial intelligence in predicting tumor response in digestive system neoplasms. From 120 studies from 2016 to 2025, applications such as endoscopy, colonoscopy, capsule endoscopy, intraoperative guidance, CT/MRI radiomics, and molecular/histology AI models were analyzed. Endoscopy achieved OR = 7.72–33.65, colonoscopy for colorectal polyps OR = 12.0 (95% CI: 10.26–14.03) and capsule endoscopy OR = 10.16 (95% CI: 8.32–12.4). Intraoperative management had OR = 8.12 (95% CI: 7.12–9.26), AI strategies for patient risk OR = 9.62 (95% CI: 7.663–11). Radiomic models for rectal/colorectal tumors had OR = 10.48 (95% CI: 9.66–11.36), for gastric/esophagogastric/esophageal cancers OR = 10.81 (95% CI: 9.89–11.82). CT/MRI recurrence/prognosis models showed OR = 10.59 (95% CI: 9.52–11.79), with moderate to high methodological quality (RQS) and low to moderate risk of bias (PROBAST). Prospective multicenter studies and standardized reporting are needed for clinical use.