Artificial Intelligence in Robotic Surgery: Enhancing Intraoperative Decision-Making and Real-Time Adaptability – A Comprehensive Literature Review

Abstract

Robot-assisted surgeries have introduced advanced levels of dexterity, 3D visualization, and tremor filtration in minimally invasive surgeries. Despite such advances, conventional robotic platforms are still highly dependent on human intervention, with limited autonomy being provided. With the advent of AI, robotics platforms have been turned into intelligent platforms that can perform real-time decision-making and respond accordingly. This review provides an overview of advancements made in AI-assisted robotic surgery since 2018 until 2025. Two key areas of discussion include real-time decision-making and adaptability. Some of the most significant AI methods discussed are computer vision (CNNs for tissues' segmentation and landmark localization), reinforcement learning for optimal trajectory determination, predictive modeling (LSTM and transformer models), and hybrid collaboration models. The current analysis is based on the use of clinical data obtained during approximately 4,200 robotic surgeries conducted in fields such as urology, gynecology, general surgery, and orthopedics. It was found out that an AI-enhanced approach enhances surgical decision-making by 18%–32%, shortens surgery time by 15%–28%, and decreases complication risks by 22%–41%. Some of the key obstacles related to the application of AI algorithms in robotic surgeries are also analyzed. In conclusion, the paper outlines some promising research trends, among which one can highlight such aspects as multimodal digital twins, edge-AI applications, and the development of interpretable AI methods.