Synthetic artificial intelligence in cardiology: from generative models to clinical applications

Area of Science:

• Biomedical research
• Cardiovascular medicine
• Artificial intelligence

Background:

• Synthetic artificial intelligence (AI) is transforming biomedical research, but its clinical applications in cardiology are underdeveloped.
• Unlike traditional AI, synthetic AI creates realistic, patient-like data (e.g., ECGs, images, virtual cohorts).
• Existing literature lacks a comprehensive overview of synthetic AI in cardiology.

Purpose of the Study:

• To provide a practice-oriented review of synthetic AI architectures and their cardiovascular applications.
• To critically synthesize core synthetic AI models, including GANs, VAEs, Diffusion Models, Transformers, Autoregressive Models, Digital Twins, and Synthetic Cohort Simulators.
• To examine the technical, ethical, regulatory, and integration challenges of implementing synthetic AI in clinical practice.

Main Methods:

• Comprehensive literature review and synthesis of synthetic AI architectures.
• Mapping of emerging cardiovascular applications for each AI model.
• Analysis of implementation barriers, including technical, ethical, and regulatory aspects.

Main Results:

• Detailed overview of key synthetic AI architectures (GANs, VAEs, Diffusion Models, etc.) and their specific uses in cardiology.
• Identification of emerging applications in areas like ECG interpretation, cardiac imaging, and disease simulation.
• Discussion of significant challenges hindering clinical adoption, such as data privacy, model validation, and workflow integration.

Conclusions:

• Synthetic AI offers significant potential to enhance diagnostic precision and optimize workflows in cardiovascular medicine.
• Cardiologists must actively engage with and guide the development and implementation of synthetic AI.
• This review serves as a call to action for researchers and clinicians to shape the future of AI in cardiology.