AI-Assisted Clinical Decision Support in Emergency Departments: Effects on Diagnostic Accuracy, Patient Throughput, and Allied Health Workflow

Abstract

Emergency Departments (EDs) face rising patient volumes, diagnostic complexity, and workflow strain—challenges amplified in low- and middle-income countries. AI-assisted clinical decision support systems (AI-CDSS) have been promoted as tools to improve diagnostic accuracy, reduce delays, and enhance team efficiency, including the work of radiographers, laboratory technologists, and triage personnel. To examine how AI-CDSS integration affects diagnostic accuracy, patient throughput, and allied health workflow efficiency in emergency departments. This review synthesises quantitative findings from 2019–2025 empirical studies evaluating AI-CDSS in ED settings. Databases searched included PubMed, Scopus, and IEEE Xplore. Outcome measures included diagnostic accuracy, turnaround time (TAT), triage precision, and workflow performance indicators. AI-CDSS increased diagnostic accuracy for acute conditions by 8–22% across studies, particularly in sepsis, stroke, and trauma imaging. Several implementations reported 12–35% reductions in ED length-of-stay, mainly attributable to faster decision-making and reduced repeat testing. Allied health workflow improved through automated alerts, structured reporting, and prioritization algorithms, reducing technologist workload by 15–28%. Concerns included algorithm bias, over-reliance, and reduced professional autonomy. Evidence suggests AI-CDSS substantively improves ED diagnostic performance and patient flow while moderately reducing allied health burden. However, ethical risks and dependency concerns require robust training, oversight, and governance frameworks, especially in resource-constrained systems.