Arrhythmias & electrophysiologyCohort study
Artificial Intelligence–Driven Electrocardiogram Screening for Asymptomatic Left Ventricular Systolic Dysfunction in the General Population
- retrospective, single-center study
Authors' conclusion
In a real-world screening population with an extremely low prevalence of LVSD, the AiTiALVSD model demonstrated high diagnostic accuracy, supporting its potential role as a rule-out screening tool for HF prevention. Prospective validation is warranted.
Study context
- Population
- Self-referred adults in a large health screening population who underwent electrocardiogram (ECG)-transthoracic echocardiography (TTE) pairs between 2011 and 2023self-referred adults, 60,711 ECG-TTE pairs
- Participants
- 40,713 adults
- Exposure
- AiTiALVSD model for identifying left ventricular systolic dysfunction (LVSD) (left ventricular ejection fraction ≤40%) on TTE, compared with established heart failure (HF) risk scores
Main finding
As reported in the abstract
LVSD cases
ECG-TTE pairs meeting the criteria for LVSD
32 cases (0.054%)
Observational data: associations, not effects.
Further findings
| Outcome | AiTiALVSD | MESA (Multi-Ethnic Study of Atherosclerosis) 5-year HF score | Pooled Cohort Equations to Prevent HF score |
|---|---|---|---|
| Diagnostic performance for LVSD | |||
| AUROC | 0.973 | 0.696 | 0.672 |
| AUPRC | 0.328 | Not reported in the abstract | Not reported in the abstract |
| sensitivity | 90.6% | Not reported in the abstract | Not reported in the abstract |
| specificity | 99.4% | Not reported in the abstract | Not reported in the abstract |
| positive predictive value | 7.7% | Not reported in the abstract | Not reported in the abstract |
| negative predictive value | 100% | Not reported in the abstract | Not reported in the abstract |
- Diagnostic performance for LVSD: The MESA score was not designed to detect prevalent LVSD and was calculated without natriuretic peptide data, which may have disadvantaged its performance in this comparison.
Simulation: resources to detect one case of LVSD
- ECGs required
- approximately 1,841
- confirmatory TTEs required
- 13
Original publication
Artificial Intelligence–Driven Electrocardiogram Screening for Asymptomatic Left Ventricular Systolic Dysfunction in the General Population. JACC Adv. Published 18 March 2026. doi:10.1016/j.jacadv.2026.102660
Summarised from the published abstract; every number is checked against it