Hypertrophic cardiomyopathy (HCM) is a common inherited cardiovascular disorder that may cause serious complications. Accurate diagnosis of HCM is essential to mitigate adverse outcomes. The electrocardiogram (ECG), combined with advancements in machine learning (ML), presents a promising alternative for HCM diagnosis and this systematic review aimed to assess ECG-ML performance. This study aims to evaluate diagnostic accuracy of ECG-ML. A search was conducted across five major electronic databases. Studies were included if they assessed ML algorithms for diagnosing HCM using ECG data. Bivariate random-effects meta-analysis was employed to pool diagnostic metrics, and subgroup analyses were performed to explore heterogeneity. A total of 21 studies were included. The pooled area under curve was 0.964 (95% CI 0.906-0.979) Sensitivity and specificity were 0.914 (95% CI 0.847-0.953) and 0.965 (95% CI 0.889-0.989), respectively. The diagnostic odds ratio (DOR) was 250.796, and the overall accuracy was 0.959 (95% CI 0.893-0.985). Heterogeneity was observed (I² > 90%). Subgroup analyses indicated variations in diagnostic performance based on ML model type, validation methods, and geographic origin. Quality concerns were found. ML models demonstrated exceptional diagnostic accuracy in identifying HCM from ECG data, showing potential as effective diagnostic tools in resource-limited settings and primary care. However, heterogeneity and quality concerns highlight the need for standardized ML development and validation. Future research should focus on addressing these limitations, exploring explainable AI methods, and conducting to ensure clinical applicability.
Łajczak et al. (Wed,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: