Artificial Intelligence‒Guided Phenotyping of Cardiomyopathies: From Detection to Surveillance

The management of cardiomyopathies relies on timely diagnosis, precise phenotyping, multimodal risk stratification, and longitudinal monitoring. These tasks require integration of high-dimensional signals from patient history, electrocardiography, imaging, biomarkers, and longitudinal health records, data streams well-suited to artificial intelligence (AI)-enabled pattern recognition and prediction. Among inherited and acquired cardiomyopathies, hypertrophic (HCM) and transthyretin amyloid cardiomyopathy (ATTR-CM) best exemplify this opportunity. Both remain systematically underdiagnosed, span a continuous phenotypic spectrum, and are increasingly actionable through disease-modifying therapies. Emerging AI approaches applied to multimodality data streams can identify latent disease, quantify its severity, and predict its trajectory. Beyond diagnosis, AI paves the path for opportunistic screening, longitudinal surveillance, risk stratification, and novel clinical trial designs. Realizing this promise requires careful attention to label quality, generalizability, false-positive burden in low-prevalence settings, and prospective evidence that the human-AI synergy improves outcomes. In this review, we use HCM and ATTR-CM as key examples to synthesize the current evidence of AI in cardiomyopathies and propose a forward-looking framework for AI-guided disease phenotyping, with emphasis on regulation, implementation, and clinical evaluation.