Artificial Intelligence Beyond Polyp Detection in GI Endoscopy: Deployment, Bias, and Workflow Integration

Artificial intelligence (AI) in gastrointestinal (GI) endoscopy is shifting from algorithm development to real-world clinical use. Beyond polyp detection, impact now depends on how systems are deployed, evaluated, and embedded into everyday practice—safely, fairly, and efficiently. This Research Topic invites Original Research, Methods, Protocols, Brief Reports, Perspectives, and Systematic Reviews that advance AI in GI endoscopy across three pillars: deployment, bias, and workflow integration.We welcome studies on:Deployment and real-world performance: prospective/pragmatic trials, registries and post-market surveillance, human-in-the-loop evaluation, cost-effectiveness, and effects on diagnostic yield, safety, and operational efficiency. We encourage work on healthcare MLOps (model updates, versioning, drift monitoring), interoperability (DICOM/HL7/FHIR), cybersecurity, and data governance.Bias, fairness, and generalizability: external validation across centers, devices, and populations; dataset shift and domain adaptation; bias auditing/mitigation; and equity analyses across demographics and comorbidities. Transparent reporting (TRIPOD-AI, CONSORT-AI, DECIDE-AI) is encouraged.Workflow and human factors: reader–AI interaction, alert design and alarm fatigue, UI/UX, team communication and handoffs, training and credentialing, medico-legal considerations, and clinician adoption. Comparative effectiveness of AI-assisted versus standard practice in real settings is of particular interest.Scope includes colonoscopy, upper GI endoscopy, capsule endoscopy, enteroscopy, EUS, ERCP, and advanced therapeutic procedures. Applications beyond detection may include characterization and risk stratification, optical biopsy, quality assurance (e.g., mucosal exposure metrics), procedural guidance, documentation automation, and workflow orchestration. Multimodal approaches (video, imaging, clinical metadata), privacy-preserving learning (e.g., federated methods), and explainability/usability studies are welcome. Topic editors: Rabindra Watson, Vijaya J, Shivaram Poigai Arunachalam, Dalton A. Norwood Source: https://www.frontiersin.org/research-topics/75120/artificial-intelligence-beyond-polyp-detection-in-gi-endoscopy-deployment-bias-and-workflow-integration