Arrhythmias & electrophysiology
Artificial intelligence in detecting left atrial appendage thrombus by transthoracic echocardiography and clinical features: the Left Atrial Thrombus on Transoesophageal Echocardiography (LATTEE) registry
Authors' conclusion
LAT-AI allows accurate prediction of LAT.
Study context
- Population
- Patients on chronic oral anticoagulation (OAC) undergoing transoesophageal echocardiography (TOE) before cardioversion or catheter ablationdevelopment n = 2827
- Comparison
- LAT-AI vs left ventricular ejection fraction (LVEF) vs CHA2DS2-VASc score for predicting LAT
- Outcome
- AUC for predicting left atrial appendage thrombus (LAT) on TOE
Methodological details13-site prospective registry, external validation
- 13-site prospective registry
- External validation
Performance
As reported in the abstract
AUC for LAT, external cohort
| Test or model | AUC |
|---|---|
| LAT-AI | 0.85 (95% CI 0.82-0.89) |
| LVEF | 0.81 (95% CI 0.76-0.86) |
| CHA2DS2-VASc | 0.69 (95% CI 0.63-0.7) |
AUC
LAT-AI0.85 (95% CI 0.82-0.89)
LVEF0.81 (95% CI 0.76-0.86)
CHA2DS2-VASc0.69 (95% CI 0.63-0.7)
- LAT-AI vs LVEF: P < 0.0001
- LAT-AI vs CHA2DS2-VASc: P < 0.0001
Further findings
LAT prevalence, external cohort
- LAT found
- 5.5%
TOE avoided with LAT-AI protocol (99% NPV threshold)
- Patients on chronic OAC who would safely avoid TOE
- 40%
Original publication
Artificial intelligence in detecting left atrial appendage thrombus by transthoracic echocardiography and clinical features: the Left Atrial Thrombus on Transoesophageal Echocardiography (LATTEE) registry. Eur Heart J. Published 15 July 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →