Artificial intelligence models demonstrated strong diagnostic performance in differentiating hypertrophic cardiomyopathy from other forms of hypertrophy, yielding AUC values between 0.89 and 0.96.
Does artificial intelligence accurately differentiate hypertrophic cardiomyopathy from physiological or other etiologies of left ventricular hypertrophy?
Artificial intelligence shows strong potential for differentiating hypertrophic cardiomyopathy from physiological hypertrophy with high accuracy (AUC 0.89-0.96), though clinical application in sports cardiology is currently limited by a lack of athlete-specific datasets and external validation.
BACKGROUND: Hypertrophic cardiomyopathy (HCM) is the most common inherited myocardial disorder and a major cause of sudden cardiac death in young adults and competitive athletes. Distinguishing HCM from exercise-induced physiological hypertrophy is clinically critical, particularly within the diagnostic "grey zone" where morphological and functional parameters overlap. Artificial intelligence (AI), including machine learning (ML) and deep learning (DL), has emerged as a methodological framework for extracting latent diagnostic information from complex cardiovascular datasets. However, the extent of its validated application in differentiating HCM from athlete's heart remains unclear. METHODS: A scoping review was conducted according to the Arksey and O'Malley methodology and PRISMA-ScR guidelines. Four electronic databases (PubMed, Scopus, Web of Science, IEEE Xplore) were searched (up to October 2025) for primary studies employing AI/ML to discriminate HCM from physiological or other etiologies of left ventricular hypertrophy. Inclusion criteria required confirmed HCM populations, use of AI analytical methods, and reporting of diagnostic performance metrics. Data extraction focused on population source, data modality, model architecture, performance indices, and presence of athlete-specific cohorts. RESULTS: Eight studies met the inclusion criteria. Only one directly compared HCM with athlete's heart, employing deformation-derived echocardiographic parameters and Support Vector Machine classification (AUC 0.93). The remaining studies examined HCM versus hypertensive or non-pathological hypertrophy using ECG-, echocardiography-, radiomics-, or video-based AI analysis, demonstrating AUC values between 0.89 and 0.96. No recent athlete-specific datasets were identified. CONCLUSIONS: AI shows strong potential to enhance the differential diagnosis of HCM through advanced imaging and ECG pattern analysis. However, the paucity of athlete-derived datasets and limited external validation significantly constrain clinical applicability in sports cardiology. Future research should prioritise explainable, multimodal models and dedicated athlete cohorts.
Deligiannis et al. (Fri,) conducted a review in Hypertrophic cardiomyopathy and physiological hypertrophy. Artificial intelligence (AI) analysis was evaluated on Diagnostic performance in differentiating HCM from physiological or other hypertrophy. Artificial intelligence models demonstrated strong diagnostic performance in differentiating hypertrophic cardiomyopathy from other forms of hypertrophy, yielding AUC values between 0.89 and 0.96.