Artificial Intelligence in Cardiovascular Diagnostics: Integration of Laboratory Biomarkers, Medical Imaging, and Molecular Data

Abstract

The largest burden of disease and mortality worldwide is due to cardiovascular disease (CVD), accounting for an estimated 18 million deaths each year. Nevertheless, the complex and heterogeneous presentation of the disease still poses an enormous challenge in terms of optimal diagnosis, prognostication of risk, and individual treatment paradigms. Given the potential of artificial intelligence and machine learning (ML) to integrate the multimodal information available in the cardiovascular realm, such as laboratory parameters, imaging, and genomic and molecular information, into a unified, highly accurate analytical system, it has the potential to revolutionize cardiovascular diagnosis. This review outlines and summarizes current approaches to the use of AI for cardiovascular diagnostics using cardiac biomarkers (conventional and newer types of cardiac biomarkers such as cardiac troponins, natriuretic peptides, inflammatory cytokines, microRNAs, and metabolomic patterns); cardiac imaging (multi-modality imaging, such as echocardiography, computed tomography, magnetic resonance imaging, and nuclear imaging); and multi-omics data analysis (genomic, transcriptomic, proteomic, and epigenetic markers). Target clinical indications for the use of coronary artery disease (CAD), heart failure (HF), arrhythmia detection and management, and acute myocardial infarction (AMI) are discussed. In addition, a role for real-time patient monitoring with wearable digital health technologies and devices is also reviewed. The high diagnostic accuracy (area under the curve routinely > 0.90 for many indications) that these AI models have achieved to date must be overcome to overcome barriers to standardization, clinical translation and validation, explainability of the algorithms, and equal access before effective AI-based precision cardiovascular medicine can be routinely integrated into clinical practice.