Top Things to Know: Artificial Intelligence to Enhance Precision Medicine in Cardio-Oncology
Published: February 24, 2025
Prepared by Mingxi Dennis Yu, MD, Director, Cardio-Oncology; Loyola University Medical Center
- Given the significant heterogeneity in cardiovascular (CV) risk among cancer patients undergoing cancer therapy, there is a need to improve risk stratification by incorporating individualized patient, cancer, and therapy-related factors. This statement examines how artificial intelligence (AI) can be utilized to enhance risk prediction models, advancing the field toward precision medicine.
- Common tools for cardiac risk stratification, including biomarkers such as B-type natriuretic peptide (BNP) and troponin, as well as imaging modalities like echocardiography and cardiac magnetic resonance imaging (MRI), exhibit varying levels of evidence for CV risk prediction in cancer patients.
- High-precision testing in multi-omics—including genomics, transcriptomics, proteomics, and metabolomics—offers insights into mechanisms of cardiotoxicity. While multi-omics is in a developmental phase in cardio-oncology, it shows significant potential for clinical impact.
- AI-driven prediction models utilizing electrocardiogram (ECG) data can identify cardiac conditions such as reduced ejection fraction, myocardial ischemia, and susceptibility to atrial fibrillation. These models are under study for predicting various forms of cardiotoxicity.
- AI has the potential to enhance the accuracy and reliability of cardiac imaging analysis for cancer patients receiving or having received cardiotoxic therapy.
- Natural language processing (NLP) techniques, such as machine learning, offer the ability to extract valuable unstructured data from electronic medical records.
- AI-driven methodologies could support drug discovery for cancer therapies, predict potential CV toxicities in future treatments, and inform therapeutic approaches for managing these toxicities.
- AI can enable the development of integrative models that combine clinical factors, biomarkers, imaging findings, and multi-omics to improve cardiotoxicity risk stratification. These models will require validation in large-scale clinical trials before widespread clinical adoption.
- The development of AI algorithms for cancer patients at risk for CV toxicity necessitates extensive, diverse, and representative datasets to reduce bias. This statement addresses current limitations and the biases inherent to AI and machine learning in the cardio-oncology patient population.
- The implementation of AI in precision medicine for cardio-oncology poses challenges regarding data security, privacy, potential biases, and ensuring diverse and equitable access. Thus, developing AI in this field will require careful oversight from key stakeholders, including clinicians, researchers, regulatory bodies, and reimbursement entities.
Citation
Khera R, Asnani AH, Krive J, Addison D, Zhu H, Vasbinder A, Fleming MR, Arnaout R, Razavi P, Okwuosa TM; on behalf of the American Heart Asso¬ciation Cardio-Oncology and Data Science and Precision Medicine Committees of the Council on Clinical Cardiology and Council on Genomic and Precision Medicine; Council on Cardiovascular Radiology and Intervention; and Council on Cardiovascular and Stroke Nursing. Artificial intelligence to enhance pre¬cision medicine in cardio-oncology: a scientific statement from the American Heart Association. Circ Genom Precis Med. Published online February 24, 2025. doi: 10.1161/ HCG.0000000000000097