Applications of Artificial Intelligence in Managing Patient Flow and Enhancing Safety in Emergency Departments in Saudi Hospitals

Abstract

The persistent challenge of overcrowding and patient safety risks in Emergency Departments (EDs) remains a critical bottleneck in healthcare delivery in Saudi Arabia, as existing operational frameworks fail to account for the interplay between technological innovation and human-centered implementation. Resolving this gap is essential for advancing healthcare digital transformation under Vision 2030 and improving the reliability of patient flow and safety protocols. This study aimed to systematically investigate the influence of an Artificial Intelligence-based Predictive Disposition and Triage Support System (AI-PDTSS) on key ED performance indicators. A sequential explanatory mixed-methods design was implemented, integrating a quasi-experimental pre-post analysis of 2,400 patient encounters with in-depth thematic analysis of 24 interviews with healthcare professionals across three tertiary care hospitals in Riyadh. Quantitative data acquisition focused on length of stay (LOS), door-to-physician time (DTP), and left-without-being-seen (LWBS) rates, with results subjected to rigorous statistical validation using independent t-tests, Mann-Whitney U, and chi-square tests. Findings indicated that AI-PDTSS implementation resulted in a significant improvement in all metrics: median LOS decreased by 28 minutes (p < 0.001), mean DTP reduced by 12.4 minutes (p < 0.001), and the LWBS rate was halved from 5.3% to 2.3% (p < 0.001). Subgroup analysis revealed the most substantial gains were for mid-acuity patients. Correlation between high system accuracy (87.2%) and these outcomes, alongside qualitative data, suggests the underlying mechanism is driven by reduced diagnostic uncertainty and earlier care planning, though mediated by factors like conditional trust and alert fatigue. These results provide a definitive evidence base for the operational efficacy of AI in Saudi EDs, demonstrating its potential to transform flow management. By reconciling quantitative outcomes with qualitative insights, this research offers a novel paradigm for sociotechnical implementation and contributes a scalable methodology for addressing ED crowding. The integration of these findings into national digital health strategies will facilitate enhanced precision and efficiency in emergency care delivery.