Abstract
BACKGROUND: AI-integrated wearable devices (WDs) offer real-time cardiovascular disease (CVD) monitoring with the potential to improve early detection and patient-centred care. However, the breadth, quality, and clinical applicability of supporting evidence remain variable. METHODS: A narrative review was conducted of peer-reviewed original studies, clinical trials, reviews, and regulatory documents published in English from January 2013 to June 2025. A structured Boolean search identified literature on AI-enabled WDs for cardiovascular monitoring. Selection criteria prioritised randomised trials, prospective observational studies, and real-world implementation reports with a clear regulatory context. Studies were narratively analysed for device classification, physiologic parameters measured, AI techniques, clinical validation outcomes, and evidence of patient engagement. RESULTS: Evidence demonstrates high diagnostic accuracy for arrhythmia detection and promising outcomes in heart failure monitoring, particularly with Food and Drug Administration or Conformité Européenne-approved devices. Nonetheless, most studies were small, short-term, and conducted in controlled settings, limiting generalisability. Key gaps include underrepresentation of diverse populations, lack of interoperability with electronic health records, limited AI explainability, and incomplete cost-effectiveness evaluation. Patient engagement was inconsistently addressed, with most reporting only basic usability or satisfaction measures rather than co-design or adherence optimisation strategies. CONCLUSION: AI-integrated WDs show considerable promise in enhancing CVD detection and management but require large-scale pragmatic trials, standardised interoperability protocols, transparent AI model interpretability, and structured patient involvement. Addressing these gaps will be essential for equitable, scalable, and clinically integrated adoption.