Key result
Greater decreases in AI-ECG LVSD scores predict a ~4% higher likelihood of recovery to HFiEF.
Why the study?
AI-ECG accurately detects LVSD, but whether serial changes in AI-ECG probability scores can forecast myocardial recovery to heart failure with improved ejection fraction remains unproven.
Do baseline and sequential 12-lead AI-ECG LVSD probability scores predict recovery to HFiEF in patients with HFrEF?
Do baseline and sequential 12-lead AI-ECG LVSD probability scores predict recovery to HFiEF in patients with HFrEF?
Serial changes in AI-ECG LVSD probability scores can non-invasively predict ejection fraction recovery in patients with HFrEF.
Answers are grounded in this paper and cite their sources.
May support serial AI-ECG monitoring for HFrEF recovery; extends prior detection models but leaves open prospective validation.
Background Artificial intelligence-enabled electrocardiography (AI-ECG) accurately detects left ventricular systolic dysfunction (LVSD), but its value in forecasting myocardial recovery—heart failure with improved ejection fraction (HFiEF)—via serial probability score changes remains unproven. Purpose To determine whether baseline and sequential 12-lead AI-ECG LVSD probability scores predict recovery to HFiEF in patients with heart failure with reduced ejection fraction (HFrEF). Methods In this single-centre, retrospective cohort study, adults (≥19 years) with echocardiography-confirmed LVSD (baseline LVEF ≤ 40%) were identified. Each ECG was paired with an echocardiogram performed within 14 days and analysed by an AI-ECG model to generate an LVSD probability score. HFiEF was defined as an absolute increase in LVEF of ≥ 10% at follow-up echocardiography ≥ 90 days after baseline. The Delta score (baseline minus follow-up AI-ECG probability) stratified patients into tertiles. Associations with HFiEF were assessed using Cox proportional hazards regression, and Kaplan–Meier analysis compared recovery across Delta score tertiles. Results Among 832 patients (mean age 64.0 ± 14.0 years, 66.6% male), 426 (51.2%) achieved HFiEF. Those who recovered were younger (62.1 ± 13.7 vs. 66.1 ± 14.0 years; p < 0.001) and less likely to have ischaemic heart disease (38.3% vs. 58.1%; p < 0.001). Baseline LVEF was lower in the HFiEF group (29.2% vs. 32.2%; p < 0.001) and rose more markedly at follow-up (49.9% vs. 33.3%; p < 0.001). (Figure 1) Baseline AI-ECG scores were comparable (57.6 vs. 53.6; p = 0.069) but declined substantially in the HFiEF group at follow-up (19.5 vs. 48.5; p < 0.001). () Each 1-point higher baseline AI-ECG score predicted a 0.9% lower chance of HFiEF (adjusted HR 0.991; p < 0.001), whereas each 1-point increase in Delta score predicted a 3.9% higher likelihood of recovery (adjusted HR 1.039; p < 0.001; interaction p = 0.004). Kaplan–Meier curves showed significant differences in HFiEF incidence across Delta score tertiles (p < 0.0001). (Figure 2) Conclusion Lower baseline AI-ECG LVSD probability scores and greater serial decreases independently predict ejection fraction recovery in HFrEF, offering a non-invasive means to identify HFiEF without exclusive reliance on serial echocardiography.
No takes yet. Share an insight, caveat, or question.
Lee et al. (2026) studied this question. Lower baseline AI-ECG scores predicted a 0.9% decrease in the chance of heart failure improvement, while greater score decreases predicted a 3.9% higher recovery likelihood.
Synapse has enriched 10 related papers. Consider them for comparative context: