Mandeep Mehra, MD, MSc, Milica Vukićević, MD, and Ameesh Isath, MBBS, from the Mass General Brigham Heart and Vascular Institute, are authors of a new article in the Journal of Cardiac Failure, Digital Twins and Artificial Intelligence in Heart Failure: The Premise and the Promise. Here are five things to know:
Heart failure is a complex condition where the heart is no longer able to pump blood effectively to meet the bodies demands. Symptoms can include shortness of breath, fatigue, swelling, weight gain and trouble sleeping.
Modern heart failure care still waits for patients to fail before anyone realizes it and starts treatment. By the time weight rises or labs cross a “critical” threshold that alerts clinicians to the worsening of the disease, the physiology has already turned.
But those thresholds are based on large groups of people, not on how one specific patient’s body normally behaves. That means warning signs can be missed.
AI-powered digital twins are like a personalized, virtual version of a patient. A true digital twin is not a risk score. It is a continuously updated physiological model built on equations of cardiac, vascular, renal and metabolic function, recalibrated every time new data arrive.
Digital twins use many types of data—heart rate patterns, sleep quality, lab results, symptoms and more—to create a constantly updated model of what’s happening inside that person’s body.
Because these digital twins learn how that patient usually functions, they can help doctors spot small changes early and even predict when trouble is coming. That allows doctors to step in sooner, before a crisis happens. This turns care from reactive (responding after something goes wrong) to proactive (preventing problems before they happen). It can also help hospitals use resources more wisely.
There are four ways that AI-powered digital twins can improve care for heart failure patients:
-
- Spotting Problems Early: Right now, heart failure is usually caught after someone starts getting worse—often when symptoms finally hit a certain “danger zone.”
A digital twin, however, constantly watches many small signals from a patient’s body—like heart rhythm changes, breathing patterns, kidney function and even sleep quality. Because it learns what’s normal for that specific person, a digital twin can warn doctors days or even weeks before a crisis happens. - Improving Clinical Trials: Clinical trials usually need large groups of people to compare treatments. With AI-powered digital twins, researchers can create a computer model showing how each patient would likely have done on standard care, providing a virtual control arm.
They can then compare that prediction to what actually happened when the patient received the new treatment.
This approach can reduce the number of people needed in a trial, speed the development of new treatments and potentially increase the diversity of trial participants.
These models are not substitutes for randomization, however. They require calibration arms, drift monitoring (continuous monitoring and updating of the twin's performance) and regulatory oversight. - A Deeper Understanding of Heart Failure Types: Doctors often classify heart failure in broad categories based on “ejection fraction” (how much blood the heart pushes out).
But this is a somewhat general approach that doesn’t capture the complexity of the disease.
Digital twins can combine many layers of information—blood flow, immune system activity, metabolism and how medical devices interact with the heart—to create a much more detailed picture of what’s going on.
This helps clinicians tailor treatments to the unique biology of each patient. - Helping Hospitals Plan Better: Heart failure care puts a heavy load on hospitals, clinics, and home‑care programs.
Right now, most planning is based on what happened in the past. Digital twins can help health systems look ahead and predict things like upcoming spikes in heart‑failure patients and when more staff or hospital beds might be needed.
Dashboards extrapolate the past. Digital twins simulate the future.
- Spotting Problems Early: Right now, heart failure is usually caught after someone starts getting worse—often when symptoms finally hit a certain “danger zone.”
To use digital twins safely and effectively in heart‑failure care, several things need to be done carefully:
-
- Make sure the model is accurate: AI powered digital twins must be thoroughly tested to prove they truly reflect how the heart works. If the model isn’t trustworthy, it could give misleading results.
- Make the tools easy for clinicians to use Even the smartest technology isn’t helpful if doctors can’t understand it quickly or apply it in real‑world care.
- Use fair, representative data If the data used to train digital twins comes mostly from one type of patient, the model may not be applicable to different presentations of the disease. Ensuring diverse, inclusive data helps avoid bias and keeps care equitable.
- Have strong oversight and transparency Health systems, regulators, and developers need clear rules and open processes so everyone understands how digital twins work, when they’re appropriate to use, and how their predictions are generated.
Heart failure has always punished imprecision," says Mehra. "Digital twins are valuable not because they are new, but because they restore something medicine has lost: Continuous physiological understanding."
Paper Cited
Mehra, M. R., Vukićević, M., & Isath, A. (2026). Digital Twins and Artificial Intelligence in Heart Failure: The Premise and the Promise. Journal of cardiac failure, S1071-9164(26)00002-3. Advance online publication. https://doi.org/10.1016/j.cardfail.2025.12.009
Subscribe to BenchMarks!
Subscribe to our monthly research newsletter and be the first to know what’s shaking in science at Mass General Brigham. From groundbreaking discoveries to the latest “you-heard-it-here-first” breakthroughs, we’ve got the updates that’ll keep your curiosity grooving.
Leave a Comment