Perioperative medicine & outcomes
Artificial intelligence‐assisted peri‐operative echocardiography: a multicentre observational study
Authors' conclusion
Agreement between US2.AI and clinicians was high for 10 echocardiographic parameters.
Study context
- Population
- Adults referred for pre-operative echocardiography202 analysed
- Participants
- 206 adults
- Exposure
- AI-derived vs clinician manual echocardiographic measurements
- Outcome
- Agreement between AI- and clinician-derived measurements for 10 key echocardiographic parameters
Methodological detailsmanual clinician measurements as reference standard
- Manual clinician measurements as reference standard
Main finding
As reported in the abstract
Agreement for continuous measurements
- Good to excellent agreement for most parameters, intraclass correlation coefficient (p < 0.001)
- 0.605-0.956
- Left ventricular ejection fraction correlation (p < 0.001)
- r = 0.845
- Left ventricular ejection fraction mean difference
- −1.9%
- Left ventricular systolic function classified correctly
- 180/201 (91%)
- Left ventricular diastolic dysfunction classified correctly
- 193/201 (96%)
- Right ventricular size, right ventricular function, right atrial size correlations (all p < 0.001), small mean differences
- r = 0.860, 0.743 and 0.842
- Inferior vena cava collapsibility: moderate agreement, low mean bias
- r = 0.641
- Cardiac output: moderate agreement, low mean bias
- r = 0.675
- Cohen's κ for categorical classifications statistically significant for all parameters
- p < 0.001
Observational data: associations, not effects.
Original publication
Artificial intelligence‐assisted peri‐operative echocardiography: a multicentre observational study. Anaesthesia. Published 5 May 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →