Natural Language Processing (NLP)- and Machine Learning (ML)-Enabled Operating Room Optimization: A Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Systematic Review Anchored in Project Planning Theory

Abstract

The operating room (OR) is a high-stakes, resource-intensive environment where inefficiencies in scheduling, workflow, and resource allocation can significantly impact patient outcomes and healthcare costs. Emerging technologies such as natural language processing (NLP) and machine learning (ML) offer data-driven solutions to optimize surgical workflows, particularly when integrated with structured project planning principles. This systematic review evaluated how NLP and ML techniques, grounded in project management methodologies, can enhance OR management by improving surgical scheduling, workflow efficiency, and resource utilization. A systematic search of PubMed, Scopus, Web of Science, Institute of Electrical and Electronics Engineers (IEEE) Xplore, and Association for Computing Machinery (ACM) Digital Library was conducted between January 1, 2020, and March 15, 2025, following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines. Inclusion criteria focused on studies applying NLP or ML to surgical workflow analysis within a project planning framework. Primary outcomes included improvements in surgical duration prediction, post-anesthesia care unit (PACU) length-of-stay estimation, and OR scheduling efficiency. Nineteen studies met the eligibility criteria, encompassing diverse surgical specialties and geographical settings. Most employed retrospective observational designs using ML models such as ensemble learning, neural networks, and regression-based algorithms. Several studies demonstrated that ML models significantly outperformed traditional scheduling and prediction approaches, while NLP, particularly ClinicalBERT, improved accuracy when analyzing unstructured clinical texts. Risk of bias assessment using the Prediction model Risk Of Bias ASsessment Tool (PROBAST) revealed that five studies were of low risk, eight moderate risk, and six high risk, primarily due to limitations in analysis and external validation. Overall, integrating NLP and ML with project planning principles presents a promising approach to optimizing OR workflows, enhancing efficiency, reducing costs, and improving patient outcomes. However, broader clinical adoption will require cross-institutional validation, improved interpretability, and ethical artificial intelligence (AI) governance.

Ähnliche Arbeiten