Artificial Intelligence Applied to Echocardiography for Detection of Heart Failure with Preserved Ejection Fraction and Elevated Filling Pressures: A Diagnostic Test Accuracy Meta-Analysis

Heart failure with preserved ejection fraction (HFpEF) accounts for nearly half of heart failure cases and remains difficult to diagnose non-invasively. Artificial intelligence (AI) applied to echocardiography may improve detection, but its aggregate diagnostic accuracy has not been quantified. We performed a diagnostic test accuracy meta-analysis of AI-enabled echocardiography for identifying HFpEF, elevated left ventricular filling pressure, or diastolic dysfunction. MEDLINE/PubMed, Embase, and CENTRAL were searched from inception to June 2026. Studies reporting the accuracy of AI models applied to echocardiography against invasive, guideline-adjudicated, or ASE/EACVI reference standards were included. Sensitivity and specificity were pooled using a bivariate random-effects (Reitsma) model, with a summary receiver operating characteristic (ROC) curve, diagnostic odds ratio, and likelihood ratios. Risk of bias was assessed with QUADAS-2. Eight studies (2770 participants; 1317 [47.5%] with the target condition) were included. Pooled sensitivity was 84.0% (95% CI: 73.0–91.0%) and specificity 83.0% (95% CI: 73.0–89.0%), with an area under the summary ROC curve of 0.875 (95% CI: 0.791–0.930). The diagnostic odds ratio was 24.73 (95% CI: 18.46–33.12); positive and negative likelihood ratios were 4.84 and 0.20. Heterogeneity was high. Performance did not differ by AI architecture but varied significantly by reference standard. No small-study effects were detected. AI-enabled echocardiography shows good pooled accuracy for detecting HFpEF and elevated filling pressures, supporting its diagnostic potential. High heterogeneity and reference-standard dependence warrant larger, prospectively validated studies before routine clinical adoption.