Evaluation of Artificial Intelligence-Generated Emergency Department Summaries and their Impact on Hospital Handoff Efficiency and Patient Outcomes

Artificial Intelligence (AI) has gained significant traction in health care, particularly in automating documentation processes. AI-generated emergency department (ED) summaries have the potential to enhance hospital handoff efficiency and improve patient outcomes. However, their accuracy, reliability, and impact on clinical workflow remain underexplored. This study evaluates AI-generated ED summaries in comparison with traditional physician-written reports concerning handoff efficiency, documentation time, and patient safety.

A prospective, randomized controlled study was conducted in a tertiary care hospital over six months. A total of 500 ED patients were included, randomly assigned into two groups: AI-generated summaries (n = 250) and physician-written summaries (n = 250). The AI model utilized for summary generation was trained using natural language processing (NLP) techniques. Key evaluation parameters included handoff completion time, documentation accuracy, physician satisfaction (measured on a 5-point Likert scale), and adverse patient events. Statistical analysis was performed using SPSS v. 26, with significance set at P < 0.05.

The AI-generated summaries significantly reduced handoff completion time (8.2 ± 1.5 min vs. 12.7 ± 2.1 min, P < 0.001). Documentation accuracy was comparable between AI-generated and traditional reports (92.4% vs. 94.1%, P = 0.07). Physician satisfaction was higher for AI-generated summaries (4.3 ± 0.6 vs. 3.7 ± 0.9, P < 0.01). Adverse events were not significantly different between the groups (2.4% vs. 2.8%, P = 0.62).

AI-generated ED summaries demonstrate significant potential in enhancing hospital handoff efficiency by reducing completion time without compromising accuracy or patient safety. The findings suggest that AI integration into clinical documentation could improve workflow efficiency and physician satisfaction. Further large-scale studies are required to validate these results across different healthcare settings.