A recent statement by the American Heart Association (AHA) on artificial intelligence (AI) in cardiovascular care examines into several key domains, offering detailed insights into the application of AI/ML (machine learning) across diverse areas, including genetics, electronic health records (EHRs), and a comprehensive framework for successful AI/ML implementation. In genetics, the AHA highlights the transformative potential of AI/ML algorithms in predicting common cardiovascular diseases. These algorithms leverage personal genomics data to enable effective preventive medicine and clinical surveillance. The statement also emphasizes the identification of monogenic causes of cardiovascular diseases, a key aspect for targeted drug development. AI/ML facilitates the development of highly efficacious novel drug therapies by discovering genes responsible for cardiovascular diseases, as demonstrated by the case of statin drugs. The statement further addresses the need for AI/ML algorithms to improve the classification of rare genetic variants as benign or pathogenic. This is particularly important in clinical genetics, where targeted genetic testing is challenged by the frequent observation of genetic variants of uncertain relevance.
Transitioning to EHRs, the AHA provides insightful best practices for optimizing AI/ML algorithms. Leveraging the largest and most well-curated EHRs emerges as a key recommendation, as it allows for the development of algorithms that are robust, validated, and applicable across diverse patient populations. The iterative improvement of EHR structures based on ongoing experiences is another important aspect highlighted by the AHA. This emphasis on continuous refinement ensures that EHRs evolve to meet the various demands of healthcare practices. Challenges in this domain include the need to guarantee the accuracy and generalizability of predictive AI tools, especially concerning the often-used American College of Cardiology/American Heart Association pooled cohort risk equation. The statement also recognizes the potential of EHR-based AI/ML algorithms to complement traditional randomized clinical trials, offering a pragmatic approach to real-world trial emulation.
The AHA’s comprehensive framework for successful AI/ML implementation in cardiovascular medicine covers a range of best practices and challenges. The triangulation of AI/ML algorithms in different datasets, facilitating data sharing, emerges as a strategic approach. Benchmarking studies against current standards, accompanied by gain and cost-effectiveness analysis, ensures the robustness of AI/ML-based precision medicine algorithms. The involvement of multidisciplinary teams, consisting of clinicians, researchers, and informatics experts, is deemed necessary for optimizing treatment outcomes. The AHA also emphasizes the importance of explainability in AI/ML algorithms to promote trust and adoption among stakeholders. Challenges in this framework involve ensuring the transferability of algorithms across different patient populations, incorporating social determinants into prediction models, managing regulatory considerations, and safeguarding at-risk communities from potential discrimination.
This comprehensive overview highlights the AHA’s commitment to advancing cardiovascular health through meticulous considerations of AI/ML applications. The AHA seeks to guide the responsible and effective integration of AI/ML technologies in cardiovascular care by addressing challenges and promoting best practices in genetics, EHRs, and a broader implementation framework.
