Artificial intelligence for operating room workflow optimization using a seven-domain prediction-to-action implementation framework
Artificial intelligence (AI) increasingly supports prediction, scheduling, optimization, and workflow analysis in operating rooms (ORs), but technical performance alone does not establish operational value. This structured narrative review and conceptual synthesis examines the prediction-to-action pathway across 20 eligible studies. Studies were coded for model output, operational action, decision owner, timing, workflow integration, operational or behavioral outcomes, and implementation maturity using an author-developed M0–M6 lens. Applications included duration and OR-time estimation, cancellation prediction, capacity planning, scheduling, demand forecasting, turnover and workflow monitoring, equipment management, patient prioritization, and implementation evaluation. Model outputs and performance metrics were reported more consistently than ownership, timing, integration, behavioral outcomes, and sustained operational effects. Fifteen studies remained at M1–M3 (retrospective development, validation, or historical simulation), while five reached M4–M6, including prospective monitoring, limited deployment, one randomized scheduling trial, and one uncontrolled bundled implementation study. The review contributes a seven-domain framework and maturity lens for assessing whether OR-AI systems progress toward accountable, timely, workflow-integrated, and outcome-measured use. These tools offer researchers, hospital leaders, informatics teams, perioperative managers, and vendors a practical audit of implementation readiness.
Authors
- Dabeluchi Chiedozie Ngwu (ORCID: https://orcid.org/0000-0001-8597-9198)
- Oluranti A. Akinyemi
- Sharfuddeen Abbas Mashi
- M-iyenebari Theophilus Odagme
Publication Details
- Journal
- Scientific Reviews
- Published
- 2026-11-19
- DOI
- https://doi.org/10.1038/s44573-026-00036-8
- Primary Topic
- Healthcare Operations and Scheduling Optimization
- Type
- article
- Field-Weighted Citation Impact
- 0.00