Medical-AI trial: agreement between ChatGPT-5 and anesthesiologists on preoperative risk class (ASA-PS) — preoperative triage (NCT07459491)
An observational study in adults scheduled for elective surgery comparing the agreement between the ASA-PS physical-status class assigned by anesthesiologists and the one generated by ChatGPT-5 from the same anonymized information. It also explores differences in lab recommendations and links to perioperative red-blood-cell (transfusion) use. 703 patients; completed.
Trial overview (primary data)
- StatusCompleted
- ConditionsArtifical Intelligence, Preoperative Evaluation
- InterventionsOTHER: No intervention (observational study)
- SponsorDamla Kaytancı Özçelik
- Target enrollment703 participants
- Period2026-01-10 〜 2026-05-01
Key points
- An observational study measuring the agreement between anesthesiologist- and ChatGPT-5-assigned ASA-PS class in adults for elective surgery.
- ASA-PS is the global standard for preoperative status but depends on clinician judgment, with inter-observer variability.
- Primary endpoint is agreement of the two classifications; secondary looks at differences in lab recommendations and links to perioperative red-blood-cell use.
- Both clinicians and the AI receive the same anonymized preoperative clinical information.
- Target 703; completed in 2026. The focus is consistency, not AI superiority.
- What is measured is not whether AI outperforms physicians but how far its classification agrees with theirs.
1Preoperative assessment as the entry point
Before surgery, the overall condition of a patient is assessed to plan anesthesia and the operation and to prepare transfusion and testing. The global reference for this is the ASA-PS classification, which grades patients from healthy to those with severe systemic disease. Because the class depends on clinician interpretation, judgments can differ between evaluators (inter-observer variability).
2Measuring agreement with anesthesiologists
This trial applies a large language model (LLM), ChatGPT-5, to that preoperative assessment and, as an observational study, measures how well the AI agrees with anesthesiologists. Per the registry summary, the same anonymized preoperative clinical information is given to both clinicians and the AI, and the agreement between the ASA-PS classes they assign is the primary endpoint.
Secondarily, it examines how the recommended preoperative labs differ between AI and clinicians, and how the assessments relate to perioperative red-blood-cell (transfusion) use. Participants are adults scheduled for elective surgery, with a target of 703; the study is at the completed stage.
3Not superiority, but agreement
What this trial measures is not whether AI outperforms physicians. It is how far the AI's classification agrees with theirs.
ASA-PS classification depends on clinician interpretation and is known to vary between assessors. The same anonymised preoperative information goes to both physician and AI, with the agreement between their classifications as the primary endpoint. Preoperative assessment is performed in volume daily, and risk classification carries through to test ordering and transfusion preparation.
Why it matters
Risk classes like ASA-PS drive resource decisions such as testing and transfusion prep. Measuring on real data how closely an LLM agrees with clinicians is a useful example of validity checking before bringing LLMs into care (this trial evaluates agreement, not efficacy).
FAQ
What is the ASA-PS classification?
Did it show AI is more accurate than anesthesiologists?
Why link it to red-blood-cell (transfusion) use?
Sources (primary)
Source: ClinicalTrials.gov (U.S. NIH/NLM, public domain). This site does not provide medical advice. Verify the latest and exact details with the official source. This site is not endorsed or certified by the NIH/NLM.
- ClinicalTrials.gov (study record, original)
- NCT ID: NCT07459491