AI-DRIVEN EVOLUTION IN SURGICAL ROBOTICS: A COMPREHENSIVE REVIEW OF ASSISTED AND AUTONOMOUS SURGERY

Le résumé fourni par la source

The increasing demand for high-quality surgical care, together with the uneven distribution of experienced surgeons and advanced medical infrastructure, has motivated growing interest in AI-enabled surgical robotics. Recent advances in machine learning, multimodal foundation models, embodied intelligence, and robotic control have created new opportunities for improving perception, navigation, planning, and task-specific automation in surgical systems. However, the current evidence base remains heterogeneous: many reported capabilities have been demonstrated only in simulation, phantom models, ex vivo experiments, animal studies, or narrowly defined clinical scenarios. Therefore, claims regarding clinical outcome improvement, scalability, and healthcare equity require careful interpretation and further validation. This paper presents a review of AI-assisted and autonomous surgical robotics. We summarize recent progress across three major technical dimensions: surgical information acquisition and scene understanding, autonomous motion synthesis and control, and simulation-based validation platforms, including emerging world-model approaches. We also review foundational concepts such as levels of surgical autonomy, surgical navigation, robotic platforms, and representative AI architectures. Particular attention is given to distinguishing demonstrated capabilities from translational potential and long-term visions of autonomous surgery. The review suggests that AI-enabled surgical systems have shown promise in selected tasks, including surgical phase recognition, instrument and tissue perception, robotic skill learning, suturing, tracking, and simulation-based policy validation. Nevertheless, high-level surgical autonomy remains at an early stage, and its clinical deployment is constrained by data scarcity, limited generalizability, insufficient standardized benchmarks, safety-critical control requirements, interpretability, regulatory uncertainty, and ethical concerns regarding responsibility and supervision. Future progress will depend not only on algorithmic innovation, but also on rigorous validation, transparent human–robot collaboration frameworks, and evidence-based assessment of clinical value.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.