Clinical Report: Utilizing eTriggers and Large Language Models in EDs
Overview
This study evaluates the effectiveness of electronic trigger tools (eTriggers) and large language models (LLMs) in identifying missed diagnostic opportunities (MODs) in emergency departments (EDs). The findings indicate that eTriggers can identify cases at risk of diagnostic error, but the yield of MODs remains low. LLMs may assist in enhancing diagnostic safety review.
Background
Missed diagnostic opportunities (MODs) are a significant concern in emergency departments, contributing to patient harm. Trigger tools, particularly eTriggers, have been developed to identify patients at increased risk for MODs, but their effectiveness in large cohorts has been limited, as evidenced by previous studies. The integration of large language models (LLMs) into this process may offer a novel approach to improve diagnostic accuracy and patient safety.
Data Highlights
This study utilized a retrospective diagnostic approach across 9 hospitals, focusing on two established ED eTrigger cohorts to evaluate the performance of LLMs in identifying MODs. The methodology involved a structured review of electronic health records (EHR) to assess missed opportunities.
Key Findings
eTriggers are designed to identify cases at increased risk for MODs but have low yields in large cohorts, as shown in prior research.
LLMs can synthesize clinical records rapidly, potentially aiding in diagnostic safety reviews.
The study compared LLMs based on sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV), with results directly supported by the data.
Model discrimination was assessed using the area under the receiver operating characteristic curve (AUC).
Concordance between LLM outputs and individual physician reviewers was analyzed.
Clinical Implications
The findings indicate that while eTriggers are useful for identifying at-risk patients, their low yield highlights the need for further evaluation of LLMs. Understanding the performance of these models can inform their potential integration into clinical workflows.
Conclusion
The study presents the potential of combining eTriggers and LLMs to enhance the identification of missed diagnostic opportunities in emergency departments.
by Clifford M. Marks, Sean Gibney, Bryan Stenson, Deesha Sarma, Cynthia Gaudet, Haadi Mombini, Thomas A. Buckley, Mario Keko, Larry A. Nathanson, Laura G. Burke, Nathan I. Shapiro, Jonathan L. Burstein, Shamai A. Grossman, Anika Parab, Alexander T. Janke, Arjun K. Manrai, Richard A. Taylor, Carlo L. Rosen, Adam Rodman, Adrian D. Haimovich
Google Trends data showed increased US search interest in vitamin A and cod liver oil during the 2025 measles outbreak, but normalized search data cannot determine actual product use, dosing behavior, or harms.
