AI Driven Advances in Gastroenterology & Hepatology

Overview

This lecture by Dr. G.V. Rao explores the integration of artificial intelligence into gastroenterology and hepatology, highlighting how AI-driven tools are transforming endoscopy, hepatology, surgical decision-making, and clinical monitoring at AIG Hospitals, Hyderabad.

Key Clinical Topics

• Limitations of natural intelligence and equation-based clinical decision-making in gastrosciences
• AI-enhanced endoscopy including colonoscopy and upper GI endoscopy
• AI-guided third-space endoscopy and POEM procedures
• Capsule endoscopy advancements including remotely maneuverable PillBot
• AI applications in liver disease, including fibrosis and fatty liver quantification
• AI-powered early sepsis detection and hospital-wide monitoring
• Pancreatic disease differentiation using AI-assisted radiology
• Cholangioscopy accuracy improvement via AI integration
• AI applications in laparoscopic and IBD surgery

Key Learning Points

• Limitations of equation-based medicine: Current predictive models in gastrosciences provide acceptable estimations for only approximately half of the patient population, with the remainder being over- or underestimated
• Omics integration: Large datasets from genomics, proteomics, and clinical research require AI to be meaningfully integrated into clinical decision-making
• AI-enhanced endoscopy: Combining white light endoscopy, narrowband imaging, and AI allows real-time demarcation of areas of interest, reducing negative biopsies and supporting novice endoscopists
• Adenoma detection: AI-integrated colonoscopy enhances polyp color visualization and provides real-time alerts, improving adenoma detection rate (ADR) during routine colonoscopy
• Endo Brain (Olympus): An AI-integrated endoscope with microscopy capability that classifies lesions as neoplastic or non-neoplastic in real time, optimizing biopsy targeting
• AI in POEM and third-space endoscopy: AI algorithms identify procedural phases, alert to vessels requiring coagulation, and delineate anatomical structures such as sling fibers, reducing morbidity in novice operators
• PillBot: A remotely maneuverable capsule endoscope controlled via joystick, with future potential for drug and clip delivery to treat small bowel pathology
• AI-derived FibroScan equivalent: An algorithm using four simple blood tests generates results comparable to FibroScan for quantifying hepatic fat and fibrosis, at significantly lower cost
• Sepsis detection portal: An AI-powered, EMR-integrated command center continuously monitors patients and alerts clinicians before clinical deterioration, resulting in a dramatic reduction in code blue events
• Pancreatic ductal adenocarcinoma (PDAC): AI enables earlier radiological detection of PDAC compared to conventional cross-sectional imaging, with high sensitivity and specificity in differentiating PDAC from other pancreatic conditions
• Cholangioscopy accuracy: AI integration improves biliary stricture differentiation accuracy from 60% to 90%, with particular clinical value in distinguishing tuberculosis from malignancy in tropical settings

Clinical Data & Study Results

• Conventional cholangioscopy biopsy accuracy: approximately 60%
• AI-integrated cholangioscopy accuracy: improves to approximately 90%
• Equation-based predictive models: yield acceptable results in only ~50% of gastroenterology patients
• India's global AI publication ranking: currently number 3 in AI-related gastroscience publications
• AI IQ benchmark cited: approximately 130+, comparable to expert-level human cognition

Practical Takeaways

• Incorporate AI-assisted colonoscopy platforms to improve adenoma detection rate and reduce missed polyps, particularly for less experienced endoscopists
• Consider AI-guided POEM algorithms as a training and safety adjunct for novice therapeutic endoscopists
• Explore AI-derived fibrosis algorithms using routine blood tests as a cost-effective alternative to FibroScan, especially in resource-limited settings
• Implement AI-powered sepsis detection systems integrated with EMR to enable earlier clinical intervention and reduce inpatient deterioration events
• In tropical and endemic regions, apply **AI-enhanced cholangioscopy