Augmenting colonoscopy with artificial intelligence and robotic assistance: a human-factors scoping review toward a task-based shared autonomy paradigm

Abstract

Abstract Colonoscopy is central to colorectal cancer prevention but remains a high-demand psychomotor and cognitive procedure, characterized by operator dependency, variable performance, and substantial physical and mental workload. Increasing evidence links procedural quality and safety to clinician-related factors, including musculoskeletal strain, fatigue, and non-technical skills. Concurrently, artificial intelligence (AI) and robotic technologies are emerging as potential tools to address these limitations. We conducted a scoping review with narrative synthesis of literature on human factors, ergonomics, AI, robotic technologies, and shared autonomy in colonoscopy and surgical robotics. Evidence synthesis: Colonoscopy performance remains limited by non-negligible miss rates and variability driven by both technical and human factors. Ergonomic constraints and cognitive workload contribute significantly to operator strain and may affect vigilance and decision-making. AI systems—including computer-aided detection (CADe), diagnosis (CADx), and quality monitoring—have demonstrated improvements in detection metrics and procedural standardisation, although concerns regarding cognitive offloading and potential deskilling persist. In parallel, emerging robotic and soft-robotic platforms aim to enhance navigation, improve procedural ergonomics, and reduce operator musculoskeletal strain, although clinical evidence remains limited. These technologies converge within a task-based shared-autonomy paradigm integrating physical, perceptual, and cognitive augmentation, with different procedural tasks operating at different levels of human–system control. A task-based shared-autonomy framework may provide a coherent model for integrating AI and robotics into colonoscopy while preserving human supervision and clinical judgement. Future developments should prioritise human-centred design, workflow integration, and multidimensional evaluation strategies to ensure safe and effective clinical implementation.