UMass Memorial Health tested a real-time AI system during live endoscopy procedures that analyzed video footage to classify bile duct narrowings as cancerous or benign, as part of the SMART-AI clinical trial.

The SMART-AI trial, led by a physician at UMass Chan Medical School and conducted using patient procedures at UMass Memorial Health, involved connecting an AI-equipped computer directly to a cholangioscope (a tiny camera used inside bile ducts). The AI analyzed the live video stream frame-by-frame, classifying areas as malignant, suspicious, benign, or uninformative in real time. In a 41-patient trial, the AI achieved 87.8% accuracy in classifying biliary strictures, compared to 67.4% for standard sampling techniques and 63.1% for visual assessment by experienced endoscopists. Results were published in April 2026 in Clinical Gastroenterology and Hepatology.