Key result
Large language models trail emergency physicians in ECG agreement with cardiologists despite high major cardiovascular events sensitivity.
Why the study?
Artificial intelligence may support ECG interpretation and MACE risk stratification in emergency departments, but the clinical utility of models like ChatGPT and Gemini remains uncertain.
Does AI-assisted ECG interpretation (ChatGPT and Gemini) accurately interpret ECGs and stratify 30-day MACE risk compared to emergency specialists and cardiologist consensus in emergency department patients presenting with chest pain or palpitations?
Cohort (n=89)
No
Does AI-assisted ECG interpretation (ChatGPT and Gemini) accurately interpret ECGs and stratify 30-day MACE risk compared to emergency specialists and cardiologist consensus in emergency department patients presenting with chest pain or palpitations?
ChatGPT and Gemini showed lower ECG interpretation agreement with cardiologists than emergency specialists, indicating that these AI models cannot currently be used to rule out pathology and are at most supplementary decision-support tools despite high sensitivity for 30-day MACE risk.
Answers are grounded in this paper and cite their sources.
May not yet substitute for cardiologist ECG review in ED chest pain; leaves open whether refined models can achieve specialist-level agreement for MACE stratification.
Background Artificial intelligence (AI) may support electrocardiogram (ECG) interpretation and risk stratification in emergency departments, but its clinical utility remains uncertain. Objective To evaluate the performance of ChatGPT and Gemini in ECG interpretation and risk stratification for major adverse cardiac events (MACE). Methods In 89 prospective ED patients presenting with chest pain or palpitations, ECGs evaluated by AI models and emergency specialists were compared against cardiologist consensus. Cohen's kappa assessed agreement; diagnostic metrics were calculated for MACE risk stratification, and McNemar's test compared paired positive-call frequencies between the models. Results Emergency specialists showed high agreement with cardiologists (κ = 0.689–0.874), whereas ChatGPT (κ = 0.386) and Gemini (κ = 0.224) demonstrated lower agreement. McNemar testing revealed Gemini had a significantly higher positive-call tendency for ischemic features (ST elevation, ST depression, T-wave inversion; p < 0.05) than ChatGPT. For 30-day MACE risk stratification (n = 27, 30.3%), ChatGPT (96.3% sensitivity, 85.4% accuracy) and Gemini (92.6% sensitivity, 79.8% accuracy) showed numerically higher performance for ChatGPT; the proportion of positive MACE risk calls did not differ significantly between models (McNemar's exact test, p = 0.55). Conclusion In this single-center pilot cohort (n = 89), both AI models showed high sensitivity in an exploratory MACE risk-stratification analysis, with no significant difference in positive-call frequency; equivalence cannot be inferred. Both showed lower agreement with cardiologists than emergency specialists. These preliminary findings do not support rule-out use and, at most, suggest a supplementary decision-support role, requiring further multi-center validation.
No takes yet. Share an insight, caveat, or question.
Doru et al. (2026) conducted a cohort in Emergency department patients presenting with chest pain or palpitations (n=89). ChatGPT and Gemini AI models for ECG interpretation vs. Emergency specialists and cardiologist consensus was evaluated on ECG interpretation agreement with cardiologist consensus and 30-day MACE risk stratification (Kappa 0.386 (ChatGPT) and 0.224 (Gemini) vs 0.689-0.874 (emergency specialists), p=0.55). ChatGPT and Gemini showed lower ECG agreement with cardiologists than emergency specialists, with 96.3% and 92.6% sensitivity for 30-day major cardiovascular events.
Synapse has enriched 10 related papers. Consider them for comparative context: