<![CDATA[Background: Congenital heart disease (CHD) is a major global health concern, and early detection is critical for preventing adverse outcomes. Cardiac auscultation, the primary screening method, suffers from high inter-observer variability and modest accuracy, leading to both missed diagnoses and unnecessary specialist referrals. Artificial intelligence (AI)-assisted phonocardiography has emerged as a promising solution, but its overall diagnostic performance has not been quantitatively synthesized. Methods: Following the PRISMA 2020 guidelines, a systematic literature search was conducted across PubMed, Cochrane CENTRAL, and Google Scholar to identify studies evaluating the diagnostic accuracy of AI-assisted phonocardiography in pediatric populations (≤18 years). Eligible studies used echocardiography as the reference standard and provided sufficient data to construct a 2x2 contingency table. The methodological quality of included studies was assessed using the QUADAS-2 tool. A bivariate random-effects model was employed to calculate pooled sensitivity, specificity, likelihood ratios, and the diagnostic odds ratio (DOR), which were visualized using forest plots and a Summary Receiver Operating Characteristic (SROC) curve. Results: The selection process yielded 16 studies for inclusion in the meta-analysis. The pooled analysis demonstrated a summary sensitivity of 91.7% (95% CI: 87.3–94.7) and a summary specificity of 95.3% (95% CI: 91.8–97.4) for the detection of pathological murmurs or CHD. The positive likelihood ratio was 19.7 (95% CI: 10.8–35.6), the negative likelihood ratio was 0.087 (95% CI: 0.055–0.136), and the DOR was 226.6 (95% CI: 89.8–571.5). Methodological quality assessment revealed that nearly half of the studies (43.8%) had a high risk of bias in the patient selection domain, primarily due to the use of case-control designs. Conclusion: AI-assisted phonocardiography exhibits high diagnostic accuracy for detecting pathological heart murmurs and CHD in children, significantly surpassing the performance of conventional auscultation. However, its effectiveness is reduced for subtle acoustic signals, and the evidence base is partly limited by methodological issues in primary studies. With further validation in large-scale prospective trials, this technology holds transformative potential as a scalable, cost-effective screening tool to improve health equity in pediatric cardiac care globally.]]>
Copyrights © 2025
