Development and preliminary evaluation of an AI-enhanced three-dimensional integrated quality model for...

Traditional quality measurement systems in operating rooms often fail to capture the interdependence among medical equipment performance, operational efficiency, and staff effectiveness. Artificial intelligence and machine learning technologies may strengthen quality-monitoring frameworks when they are embedded within clearly defined operational protocols and human workflow support.

To develop and preliminarily evaluate an artificial intelligence-enhanced three-dimensional integrated quality model for quality-sensitive indicators that integrates real-time equipment monitoring, predictive analytics, and staff performance metrics within an operating room (OR) quality-management intervention.

A prospective single-center study was conducted at Sir Run Run Shaw Hospital, Zhejiang University School of Medicine, Hangzhou, China (n = 347 equipment units, 156 OR staff) from January 2024 to December 2024. The study used a pragmatic pre-post implementation design in which an AI-enhanced three-dimensional integrated quality model was implemented together with structured staff training, protocol reinforcement, and workflow optimization. Primary outcomes included equipment utilization efficiency, operational performance metrics, and staff productivity indices measured through IoT sensors, electronic health records, and validated assessment tools, with analyses specified at the equipment, room, staff, or system level according to the outcome.

Implementation of the integrated AI-enhanced model was associated with significant improvements across measured domains. Equipment downtime decreased by 37.1% (mean decrease 4.6 events/month, 95% CI: 3.8-5.4, Cohen's d = 1.52, raw and FDR-adjusted p < 0.001), staff efficiency index increased by 25.3% (mean increase 18.5 points, 95% CI: 16.2-20.8, Cohen's d = 2.41, raw and FDR-adjusted p < 0.001), and room turnover time decreased by 33.1% (mean decrease 9.5 min, 95% CI: 7.2-11.8, Cohen's d = 1.89, raw and FDR-adjusted p < 0.001). The predictive failure algorithm achieved 92.8% accuracy (95% CI: 89.4-95.3) with excellent discrimination (AUC-ROC = 0.94, 95% CI: 0.91-0.97) and calibration (Brier score = 0.082). Multivariate analysis identified maintenance protocol adherence (beta = 0.73, raw and FDR-adjusted p < 0.001) and technology integration scores (beta = 0.58, raw and FDR-adjusted p < 0.001) as the strongest predictors of system performance [R ² = 0.847; adjusted R ² = 0.842; F (10,336) = 186.0, p < 0.001].

The AI-enhanced three-dimensional integrated quality model may offer a structured framework for comprehensive OR quality management when combined with evidence-based maintenance protocols, staff training, and workflow redesign. Given the single-center pre-post design and concurrent implementation components, the findings should be interpreted as improvements observed during an integrated quality-management intervention rather than as the isolated causal effect of AI alone. Controlled multicenter studies are needed to quantify the independent contribution, transferability, and long-term sustainability of the AI components.