Anne van Dalen

260 I Summary and General Discussion In Chapter 8 the use of artificial intelligence (AI) applications in the OR is systematically reviewed. Pubmed, Embase, Cochrane library, and IEEE Xplore were searched and this yielded 193 articles. Finally, 9 studies were included. The identified applications of AI in the OR included; procedure duration prediction, gesture recognition, intra-operative cancer detection, intra-operative video analysis, workflow and phase recognition, human detection and pose estimation, an endoscopic guidance system, knot-tying, and automatic registration, and tracking of the bone in orthopaedic surgery. The great majority, if not all, of the AI applications have shown superior results in comparison to their non-AI alternatives. However, studies are set up in various pilot settings and only a small minority has been able to situate their impacts and associated changes in current health systems. According to Rogers’ widely-used Diffusion of Innovations theory, adoption of innovative technology always involves early and late adopters. During the innovation process, where an individual is motivated to reduce uncertainty about the advantages and disadvantages of an innovation, it is important to emphasize the ethical and legal challenges. 46 59 Yet, sufficient political, regulatory, organizational, and clinical conditions for AI development and ethical use of sensitive information are still lacking, and hence needed to implement AI applications safely and sustainably in the future. 60-62 Additional barriers for the widespread implementation of AI in healthcare may be unawareness on the topic or solutions, lack of user or implementation knowledge by the healthcare professionals and their workplace supporters, unresolved questions about ethics or privacy from management, or an insufficient IT infrastructure. Most likely, it will be a combination of these barriers. 63 Indeed, the application and implementation of AI inside the OR still has several challenges to overcome. However, in the not so distant future, evolving technology like the ORBB, with integrated AI and machine software, may prove to be of great help in analysing and optimizing workflow and outcome in real- time. The ORBB team is currently working on developing applications of AI in surgical practice: identifying intraoperative events like bleeding, hypothermia, aberrant anatomy, or tool use, using patient, team, and surgeon factors. The team’s attention can then be drawn to warnings and they can decide whether to take action based on the by the ORBB provided explanations: to modify risk factors by, for instance, reviewing imaging,