Title : The integration of AI algorithms for autonomous decision-making during orthopaedic surgery
Abstract:
Artificial Intelligence (AI) is increasingly integrated into orthopedic surgery, with applications across diagnosis, preoperative planning, intraoperative guidance, postoperative monitoring, and rehabilitation. This literature review assesses the implementation of AI algorithms for autonomous decision-making in orthopedic surgery, focusing on technological developments, clinical applications, progression from assistive systems to semi-autonomous and fully autonomous systems, and considerations of cost, feasibility, ethics, and regulation.
A comprehensive literature search was conducted using PubMed/MEDLINE, Google Scholar, Scopus, Cochrane, and Web of Science. Peer-reviewed English-language studies published from 2015–2026 were considered, including studies of AI-based decision-making, autonomous or semi-autonomous surgical systems, robotic-assisted orthopedic procedures, imaging-based AI applications, clinical outcomes, and feasibility. Twenty-nine studies were included in the review.
Machine learning, deep learning, convolutional neural networks, computer vision, and deep reinforcement learning have been applied to fracture detection, osteoarthritis grading, imaging interpretation, surgical planning, implant positioning, and real-time intraoperative support. Deep learning has demonstrated strong performance in orthopedic imaging, while deep reinforcement learning has been proposed for automated femoral osteotomy planning. Most current orthopedic applications remain AI-assisted or semi-autonomous, with surgeons retaining a supervisory role.
AI-supported systems can improve decision-making, surgical planning, precision, repeatability, alignment, and patient outcomes. However, implementation is limited by high costs, maintenance requirements, data quality and generalizability, transparency of “black box” models, ethical and regulatory concerns, liability, automation bias, and the risk of reduced surgical skills through over-reliance on automated systems. Explainable AI and shared-control training are identified as important approaches for maintaining surgeon understanding and the ability to override systems when required. Overall, fully autonomous surgery remains distant; the most beneficial model described is human–AI collaboration, combining computational support with the essential human role in surgical care.
