As a cardiologist specializing in cardio-oncology, I witness daily how lifesaving cancer treatments can paradoxically damage the heart. This emerging field focuses on preventing, monitoring, and treating cardiovascular complications in cancer survivors—a growing population thanks to remarkable advances in oncology.
However, cardio-oncology presents unique challenges that artificial intelligence may help us overcome. Recently, I chaired an American Heart Association scientific statement on this very topic, exploring how AI could transform our approach to these complex patients.
The fundamental challenge in cardio-oncology is the extraordinary variability between patients. Consider a typical scenario: a breast cancer patient receives a combination of therapies—chemotherapy, immunotherapy, targeted treatments, perhaps radiation. She may have pre-existing cardiovascular risk factors like diabetes or hypertension. She brings her unique genetic profile, age, and medical history.
How do we predict her specific cardiac risk? How do we personalize her surveillance and prevention strategies? Traditional research methods struggle with this complexity. Randomized controlled trials typically study narrowly defined populations under specific conditions. But our patients, each with their own unique combination of cardiovascular and oncologic risk profiles, don’t fit neatly into any these categories.
AI’s Potential for Personalized Care
This is where AI offers tremendous potential. Machine learning algorithms can analyze vast datasets encompassing
diverse patient characteristics, treatment combinations, and outcomes. Deep learning is a more specialized form of machine learning that are designed to mimic the human brain in learning and processing data, and finding and refining features within data. These systems can identify subtle patterns beyond human recognition capability, potentially allowing us to predict which patients face the highest cardiac risk from specific cancer treatments.
Imagine a future where we input a patient’s comprehensive profile—cancer type, planned treatments, pre-existing conditions, genetic information, and biomarkers—and receive personalized risk profile predicting future cardiovascular events and morbidity from a specific oncologic therapy or combination, with recommendations for cardiac monitoring and preventive strategies. This precision approach could help us move from reactive to proactive cardio-oncology care.
Addressing Data Gaps and Health Disparities
However, realizing this vision requires overcoming significant challenges, particularly regarding data quality and representativeness. Most medical research historically focused on white male populations, creating significant data gaps for women and people of color. If we build AI systems on these biased datasets, we risk exacerbating health disparities rather than reducing them.
This concern is particularly relevant for Black patients, who already face disproportionate cancer and cardiovascular disease burdens. AI systems trained primarily on data from white patients may miss crucial differences in disease presentation, treatment response, or risk factor profiles in Black populations.
The Need for a Representative Cardio-Oncology Database
That’s why I believe creating a nationally representative cardio-oncology database is essential. Rather than relying on data from a few institutions—which tend to serve relatively homogeneous populations—we need collaborative efforts spanning diverse patients across diverse healthcare settings across the country.
The American Heart Association is laying the groundwork for this important work, starting with a pilot involving several institutions. Our hope is to gradually expand these efforts to create a truly inclusive database that is regularly updated to manage potential data and concept drift where AI is concerned, and that captures the experiences of patients across racial, ethnic, and socioeconomic spectrums.
Community engagement will be crucial to this effort. Organizations serving Black communities, like BlackDoctor.org, can play vital roles in increasing research participation and ensuring AI development addresses the needs of historically marginalized populations.
Ensuring Equitable Access to AI Tools
We must also ensure cardio-oncology AI tools are designed for clinical settings with varying resources. Advanced algorithms should benefit patients whether they receive care at major academic centers or community hospitals serving underserved areas.
The integration of AI into cardio-oncology represents more than technological advancement—it’s an opportunity to deliver more equitable, personalized care. By thoughtfully building diverse, representative datasets and developing systems that account for the full spectrum of patient experiences, we can harness AI’s power while avoiding its potential pitfalls.
Cancer survivors deserve cardiovascular care tailored to their unique circumstances. With careful development and implementation, AI can help us deliver precisely that—not by replacing clinical judgment but by enhancing our ability to see each patient in their full complexity and provide truly personalized care.
Dr. Tochukwu Okwuosa is a Professor and Director of the Cardio-Oncology Program at Rush University Medical Center in Chicago, IL. She earned her medical degree from the Philadelphia College of Osteopathic Medicine and completed her Internal Medicine and Cardiology training at the University of Chicago. Dr. Okwuosa’s primary research interests lie in the areas of Cardio-Oncology and Cardiovascular Disease Prevention. She actively participates in multiple locoregional and national Cardiology/Cardio-Oncology committees and boards, serves as an Associate Editor for the Journal of the American Heart Association, and is the Immediate Past Chair of the American Heart Association’s Cardio-Oncology sub-committee.
