Cardiovascular care with digital twin technology in the era of generative artificial intelligence

Abstract

Digital twins, which are in silico replications of an individual and its environment, have advanced clinical decision-making and prognostication in cardiovascular medicine. The technology enables personalized simulations of clinical scenarios, prediction of disease risk, and strategies for clinical trial augmentation. Current applications of cardiovascular digital twins have integrated multi-modal data into mechanistic and statistical models to build physiologically accurate cardiac replicas to enhance disease phenotyping, enrich diagnostic workflows, and optimize procedural planning. Digital twin technology is rapidly evolving in the setting of newly available data modalities and advances in generative artificial intelligence, enabling dynamic and comprehensive simulations unique to an individual. These twins fuse physiologic, environmental, and healthcare data into machine learning and generative models to build real-time patient predictions that can model interactions with the clinical environment to accelerate personalized patient care. This review summarizes digital twins in cardiovascular medicine and their potential future applications by incorporating new personalized data modalities. It examines the technical advances in deep learning and generative artificial intelligence that broaden the scope and predictive power of digital twins. Finally, it highlights the individual and societal challenges as well as ethical considerations that are essential to realizing the future vision of incorporating cardiology digital twins into personalized cardiovascular care.

Keywords: Digital twins; Generative artificial intelligence; Multi-modal models; Precision medicine.

Graphical Abstract

An overview of elements comprising current digital twins in cardiovascular medicine (left panel), elements needed to build future digital twins in cardiovascular medicine (right panel), and key elements for implementation of the digital twin in healthcare (center). AI, artificial intelligence; CABG, coronary artery bypass graft; EHR, electronic health record.

Figure 1

A clinical vignette. A step-by-step clinical vignette modelling the future of healthcare with digital twins. HCTZ, hydrochlorothiazide; SGLT2, sodium–glucose cotransporter 2

Figure 2

Potential cardiovascular clinical applications of digital twins. An overview of data available and potential digital twin applications. ECG, electrocardiogram; EHR, electronic health record, ‘Omics: includes genomics, proteomics, and metabolomics data

Figure 3

Selected precision medicine applications of digital twins. (A) An example digital twins for procedural planning of atrial fibrillation ablation. 3D anatomical and electrophysiological models are derived from magnetic resonance imaging, magnetic resonance angiography imaging, and electrocardiograms to build a patient-specific digital twin. Different ablation locations are simulated, and the one most likely to stop atrial fibrillation is chosen. (B) An example simulation of a myocardial infarct scar. To substitute late-gadolinium enhancement magnetic resonance images, non-contrast magnetic resonance images and electrocardiograms can be combined to make an anatomical and electrophysiological cardiac digital twin. The digital twin can then be input into a variational autoencoder, which can generate simulations of 3D scarring, e.g. from myocardial infarction. ECG, electrocardiogram; CNN, convolutional neural network; MRA, magnetic resonance angiography; MRI, magnetic resonance imaging

Figure 4

Methodological strategies for developing digital twins. An overview of mechanistic and statistical models used in cardiac digital twins

Figure 5

Applications of digital twins for evidence generation. An overview of digital twin applications for evidence generation across clinical trial populations, for generating cardiac data, for estimating treatment effects of a clinical trial in a second patient population, and for safely testing machine learning models across multiple sites