A systematic review and meta-analysis of artificial intelligence ECGs performance in the diagnosis of Brugada Syndrome.

Abstract

Background: Brugada syndrome (BrS) is a serious condition linked to sudden cardiac death in individuals who are otherwise healthy. Notably, drug-induced BrS accounts for 50% to 70% of all documented cases. The utilization of artificial intelligence (AI) models in the analysis of electrocardiograms (ECGs) represents a promising approach for the detection of BrS.

Purpose: This meta-analysis aims to evaluate the effectiveness of AI models in diagnosing BrS through ECG analysis.

Methods: We conducted a systematic search across PubMed, Embase, and Cochrane databases, focusing on AI-based models for ECG analysis related to BrS detection. Key outcomes measured included sensitivity, specificity, and the summary receiver operating characteristic (SROC) curve. Pooled proportions were calculated using a random-effects model with 95% confidence intervals (CIs), and heterogeneity was using Zhou and Dendukuri I² approach. Additionally, a leave-one-out sensitivity analysis was performed to evaluate the impact of each one of the included studies on the pooled results and heterogeneity. All statistical analyses were conducted using R version 4.4.2.

Results: Our analysis included six studies encompassing ECG data from 2,179 patients, all employing AI algorithms for ECG interpretation. The quantitative analysis revealed an area under the curve (AUC) of 0.898, a sensitivity of 78.9% (95% CI: 69.6 to 85.9), and a specificity of 87.7% (95% CI: 79.9 to 92.7). Notably, the sensitivity analysis without Zanchi et al., significantly reduced the heterogeneity (I² = 0%). However, the other analyses corroborated with our general findings.

Conclusion: AI-driven ECG interpretation demonstrates to be a viable option in detecting BrS.