Perioperative medicine & outcomes
Comparative performance of GPT-4 models and expert anaesthesiologists in peri-operative antithrombotic management: A cross-sectional evaluation
Authors' conclusion
In simulated peri-operative antithrombotic management scenarios, a domain-specific, transformer-based language model generated faster and more complete responses than clinician-generated answers, while accuracy was comparable across groups.
Study context
- Population
- 25 hypothetical peri-operative scenarios involving anticoagulant or antiplatelet therapy, Spain25 scenarios
- Exposure
- Domain-specific model vs Anaesthesiologists
- Outcome
- Completeness and accuracy of responses, rated by three blinded experts
Methodological detailsanalytical, fully simulated, nonclinical environment, multicentre (five university-affiliated anaesthesiology departments), blinded expert rating
- Analytical
- Fully simulated, nonclinical environment
- Multicentre (five university-affiliated anaesthesiology departments)
- Blinded expert rating
Main finding
As reported in the abstract
Completeness
Domain-specific model
4.52
mean · SD 0.39
Anaesthesiologists
3.72
SD 0.43
General-purpose model
4.25
SD 0.48
P < 0.001
Observational data: associations, not effects.
Further findings
| Outcome | Domain-specific model |
|---|---|
| Accuracy | mean 4.31 (SD 0.57) |
| Incomplete and inaccurate responses | |
| Inaccurate answers | 1.3% |
- Accuracy — Accuracy did not differ significantly between groups; highest mean for the domain-specific model
- P = 0.107
Inter-rater reliability
- Inter-rater reliability range
- 0.717 to 0.885
Incomplete and inaccurate responses
- Domain-specific model produced no incomplete responses
Parts of this Brief are withheld: the automatic checks could not confirm them against the abstract.
Original publication
Comparative performance of GPT-4 models and expert anaesthesiologists in peri-operative antithrombotic management: A cross-sectional evaluation. Eur J Anaesthesiol. Published 4 May 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →