Research Appraisals
Evidence-based critical appraisals of the latest medical research, systematically evaluated using Oxford CEBM methodology.
Showing 2 appraisals
Journal of the American College of Surgeons
Improving Trauma Triage Accuracy with Large Language Models: A Comparison to Human Expert Decisions
BACKGROUND: Accurate prehospital trauma triage and communication determine morbidity, mortality, and system efficiency. Advancements in large language models (LLMs) offer an opportunity to improve triage and yet to be implemented in prehospital trauma triage. STUDY DESIGN: This retrospective cohort study evaluates LLM performance in trauma triage and accuracy of prehospital tele-communication. Of 410 pediatric activations at a Level I center (January 2023 to May 2025, IRB no. 00009569), 133 activations with emergency medical service recordings, human-generated trauma pages, and injury severity scores were analyzed. Audio was transcribed with OpenAI Whisper. Structured "Essential Transcripts" were generated with named entity recognition. Entity ablation tested redaction of triage parameters on accuracy. In a prospective arm, trauma surgeons reviewed emergency medical service transcripts and triaged activations pre- and post-LLM exposure. Cribari criteria defined over and undertriage. McNemar's test and 95% CIs assessed paired differences in accuracy. RESULTS: The primary endpoint: LLM undertriage demonstrated modest improvement; 4.8% (3 to 8.2) vs 5.1% (3.1 to 9.3, p = 0.73, Bonferroni p = 1). For secondary endpoints, LLM triage outperformed human clinicians: 83.5% accuracy (80.6 to 90.6) vs 78.9% (73.9 to 82.6, p < 0.01, Bonferroni p < 0.01), with overtriage 58.6% (51.4 to 73.7) vs 71.8% (p < 0.05, Bonferroni p < 0.09). "Essential Transcripts" reduced transcript length by 80.8% (p < 0.001) while preserving accuracy (81.9%, 76.6 to 87.5, p < 0.001, Bonferroni p < 0.05). Entity ablation had marginal effect on triage. In prospective evaluation, human triage accuracy improved after LLM exposure (73.7% [69.8 to 77.2]) to 75.8% ([71.9 to 79.2], p = 0.04, Bonferroni p = 0.12), significantly improving the odds of a correct triage decision (odds ratio 2.57, 95% CI 1.39 to 6.83, p < 0.01). CONCLUSIONS: LLMs achieve triage accuracy comparable to trauma staff in retrospective review of pediatric trauma. Further validation is needed to assess clinical outcomes, generalizability, and user acceptance before widespread deployment.
14 June 2026
Read appraisal →PLoS medicine
Machine learning assisted differentiation of low acuity patients at dispatch: The MADLAD randomized controlled trial
BACKGROUND: Resource Constrained Situations (RCS) at Emergency Medical Dispatch centers where there are more patients requiring an ambulance than there are available ambulances are common. Machine Learning (ML) techniques offer a promising but largely untested approach to assessing relative risks among these patients. The study aims to establish whether the provision of ML-based risk scores predicting patient outcomes improves the ability of dispatchers to identify patients at high risk for deterioration in RCS and dispatch the first available ambulance to them. METHODS AND FINDINGS: We performed a parallel-group, randomized trial of adult patients assessed by a dispatch nurse at two study sites in Sweden as requiring a low-priority ambulance response in RCS. Patients were randomized 1:1 to be prioritized with the aid of an ML-based risk assessment tool, or per current clinical practice. The primary outcome was defined in terms of whether the first available ambulance was sent to the patient with the highest National Early Warning Score (NEWS 2) based on subsequently collected vital signs. A total of 1,245 RCS were included in the study. In the intervention arm, 68.3% of RCS were assessed correctly per the primary outcome versus 62.5% in the control group, corresponding to an odds ratio of 1.28 (95% CI [1.00, 1.63], p = 0.047). This study was limited to only patients determined to require a low-priority ambulance response in two Swedish regions, and was underpowered for the primary outcome due to a smaller than expected sample size. CONCLUSION: This study suggests that clinical ML-based decision support tools may have the ability to influence care provider decisions and improve their capacity to rapidly differentiate between high- and low-risk patients at dispatch. Further research should establish the suitability of these tools in larger cohorts, for patients with both higher- and lower-levels of priority, and in other settings. The trial was registered at ClinicalTrials.gov (NCT04757194).
4 Apr 2026
Read appraisal →