Diagnostic Performance of Artificial Intelligence-Assisted Echocardiography in Identifying Hypertrophic...

Hypertrophic cardiomyopathy (HCM), the most common genetic cardiac disease, remains underdiagnosed most of the time due to overlapping echocardiographic characteristics and subjective interpretations. This systematic review and meta-analysis aimed to assess the diagnostic performance of artificial intelligence (AI)-assisted echocardiography interpretations for identifying HCM and to explore factors contributing to variability and validity. After a comprehensive search through various databases, eligible studies reporting diagnostic metrics such as sensitivity, specificity, or area under the curve (AUC) were included into our analyses. Data were pooled using a bivariate random-effects model, and heterogeneity was quantified with the I2 statistic. Twenty-five studies were included into our meta-analysis. The pooled AUC for AI-based echocardiographic detection of HCM was 0.93 [95% confidence interval (CI), 0.90-0.95]. After trim-and-fill correction, the pooled AUC increased to 0.96 (95% CI, 0.93-0.97). Overall sensitivity and specificity were 0.89 (95% CI, 0.83-0.93) and 0.87 (95% CI, 0.76-0.94), respectively. Meta-regression revealed that convolutional neural network, support vector machine, and ensemble learning algorithms exhibited variable performance, with convolutional neural network-based models favoring higher sensitivity. We demonstrated that AI-based models evaluating echocardiographic data could be an accurate diagnostic tool for HCM. This highlights the potential of recent advancements to improve clinical decision-making.