Acute cardiac care & resuscitation
Artificial intelligence versus human expertise: Electrocardiogram (ECG)-based detection of occlusive myocardial infarction after cardiac arrest
Authors' conclusion
QoH, despite operating without awareness of the CA-setting and thus likely at a relative disadvantage, and HE showed robust diagnostic accuracy. Due to undifferentiated overdiagnosis, general LLMs remain unsuitable for ECG interpretation.
Study context
- Population
- Adults resuscitated from cardiac arrest, post-ROSC ECG
- Participants
- 97 adults
- Comparison
- Queen of Hearts (QoH) vs human experts vs large language model (LLM)-based artificial intelligence (AI) chatbots for ECG detection of occlusive myocardial infarction (OMI) after cardiac arrest
- Outcome
- AUROC for presence and probability of OMI and acute coronary occlusion (ACO), determined by coronary angiography
Methodological detailssingle-centre
- Single-centre
Performance
As reported in the abstract
AUROC
| Test or model | AUROC acute coronary occlusion (ACO) (TIMI 0) | AUROC OMI |
|---|---|---|
| QoH | 0.846 [0.752-0.939] | 0.745 [0.647-0.843] |
| Human experts (HE) | 0.735 [0.622-0.848] | 0.635 [0.515-0.755] |
| ChatGPT | 0.456 [0.319-0.592] | 0.495 [0.376-0.614] |
| EKG Analyst | 0.474 [0.346-0.603] | 0.626 [0.508-0.743] |
AUROC acute coronary occlusion (ACO) (TIMI 0)
QoH0.846 [0.752-0.939]
Human experts (HE)0.735 [0.622-0.848]
ChatGPT0.456 [0.319-0.592]
EKG Analyst0.474 [0.346-0.603]
AUROC OMI
QoH0.745 [0.647-0.843]
Human experts (HE)0.635 [0.515-0.755]
ChatGPT0.495 [0.376-0.614]
EKG Analyst0.626 [0.508-0.743]
Further findings
Threshold-dependent performance (sensitivity/specificity)
- Both AI chatbots: high sensitivity at the cost of minimal specificity (ACO / OMI sensitivity)
- 100 % / 98.36 %
- QoH and HE: more even distributions of sensitivity/specificity
Original publication
Artificial intelligence versus human expertise: ECG-based detection of occlusive myocardial infarction after cardiac arrest. Resuscitation. Published 19 November 2025. doi:10.
Summarised from the published abstract; every number is checked against it