Artificial intelligence in cardiology
By analyzing data from monitoring devices and routine tests, this technology holds the promise of uncovering serious heart problems sooner.
- Reviewed by Christopher P. Cannon, MD, Editor in Chief, Harvard Heart Letter; Editorial Advisory Board Member, Harvard Health Publishing
How can artificial intelligence (A.I.) improve how doctors prevent, detect, and treat heart disease? The American Heart Association's first-ever scientific statement on this topic explores the current limitations and future promise of the burgeoning technology.
Harvard Medical School professor Dr. Jagmeet Singh, author of Future Care: Sensors, Artificial Intelligence, and the Reinvention of Medicine, served as a reviewer for the statement, which was published April 2, 2024, in Circulation. As a cardiac electrophysiologist specializing in A.I., digital health, and medical device technology, he's optimistic that A.I. tools can improve how doctors care for people with heart disease (see "Understanding artificial intelligence"). "But for that to happen, we'll need to disseminate this technology equitably, as we learn cost-effective strategies during the transition to new models of caring for patients," he says.
Understanding artificial intelligenceBroadly speaking, A.I. refers to systems or computers capable of mimicking human thought to solve problems. Machine learning is a type of A.I. in which algorithms learn from large amounts of data. By recognizing patterns, the machine becomes more accurate over time. Deep learning is a type of machine learning using many layered neural networks, which are modeled after the structure and function of the brain. Many modern A.I. tools are based on deep learning. Predictive A.I. harnesses data to predict the future. In medical applications, these tools aren't intended to replace the role and clinical judgment of physicians but rather to help them identify potentially worrisome conditions that are often hard to detect before concerning symptoms arise. Generative A.I. generates novel content; ChatGPT is a one example. In health care settings, this technology can streamline "back office" tasks such as scheduling, billing, summarizing notes from a visit, and answering questions people send via online medical portals. |
Faster, more accurate monitoring?
Earlier forms of A.I. have been used for years in implantable cardioverter-defibrillators (ICDs), which rely on the ability of the machine to recognize life-threatening arrhythmias (abnormal heart rhythms) and then deliver a shock to restore the heart's normal rhythm. "It's a specific form of intelligence, embedded within a device, that works much faster than a human," says Dr. Singh. Now, the algorithms have advanced to the extent that recordings of the heart's electrical activity from within an ICD (called an electrogram) may be able to predict arrhythmias well before they occur. This could enable doctors to adjust a person's treatment to prevent the device from having to deliver a shock. Other specialized sensors (which may be part of an ICD) or implantable recorders (placed near the heart) can capture additional information and accurately predict whether a person will develop heart failure within the next 30 days, Dr. Singh says.
External devices that resemble oversized Band-Aids placed on the chest may be more feasible and just as effective. Known as patch monitors, they continuously monitor the heart's electrical activity (producing an electrocardiogram, or ECG), in addition to potentially tracking temperature, oxygen level, physical activity, and fluid buildup in the lungs for up to one month. "Using A.I.-based algorithms on just 24 hours of data from a patch monitor, we can predict which patients will develop potentially serious heart arrhythmias, including atrial fibrillation or ventricular tachycardia, over the following 13 days," says Dr. Singh.
Gleaning more information from basic tests
Routine, brief ECGs done in medical settings can help diagnose arrhythmias and heart attacks. But when analyzed by A.I., that simple test result can also predict a person's ejection fraction (a measure of the heart's pumping ability, which is used to diagnose heart failure) or identify structural heart problems, such as a thickened heart muscle (a condition known as hypertrophic cardiomyopathy).
Similarly, A.I. is now capable of mining data from a single chest x-ray (one of the most commonly done medical tests) to predict a person's risk of heart attack, stroke, diabetes, and other serious cardiovascular problems.
Moving forward with A.I.
How might A.I. tools actually improve someone's health care experience? "Let's say you've recently been hospitalized with heart failure and the physician sends you home with a patch monitor and a $300 tablet computer," says Dr. Singh. With A.I. assistance, those tools allow physicians to provide virtual home-based care and remotely monitor your condition, enabling them to quickly determine if you need a medication adjustment to prevent a relapse and another costly hospitalization, he says.
Shifting A.I. monitoring efforts even earlier could have even more impact. For example, consumer products like smart watches that show people relevant data about their health and then suggest self-management strategies might improve heart disease prevention. "Personally, I think the only way to make health care more sustainable is for patients to have the tools to be more involved in their own care," says Dr. Singh.
Image: © ArtemisDiana/Getty Images
About the Author
Julie Corliss, Executive Editor, Harvard Heart Letter
About the Reviewer
Christopher P. Cannon, MD, Editor in Chief, Harvard Heart Letter; Editorial Advisory Board Member, Harvard Health Publishing
Disclaimer:
As a service to our readers, Harvard Health Publishing provides access to our library of archived content. Please note the date of last review or update on all articles.
No content on this site, regardless of date, should ever be used as a substitute for direct medical advice from your doctor or other qualified clinician.
