A new generation of artificial intelligence (AI) tools has the potential to save more patients in need of heart transplants by making better use of donor hearts that are currently discarded, according to research presented today by Brian Weida, MD, at the 46th Annual Meeting and Scientific Sessions of the International Society for Heart and Lung Transplantation (ISHLT).
“There is a significant shortage of heart donors in the United States, and patients are waiting months or more for transplants, often requiring life support in the ICU, so the risks are very high,” said Dr. Weida, an assistant professor at New York University’s Grossman School of Medicine and a heart failure and transplant cardiologist.
AI tools help surgeons make complex decisions
Recently developed AI tools are designed to help transplant teams make complex decisions about accepting or rejecting donor hearts in a more data-driven, consistent, and efficient manner. Despite the heart shortage in the United States, only about 30 to 40 percent of available hearts are actually used for transplants. Research shows that not all donor hearts are legitimately discarded.
Once a heart is available, it typically takes cardiologists and surgeons just 15 to 30 minutes to determine whether it is a good fit for a particular patient, considering a myriad of factors, including the donor’s medical history, imaging studies, and laboratory tests.
It is a very complex judgment decision that must be made in a very short time frame, often in the middle of the night. ”
Brian Weida, Assistant Professor, New York University Grossman School of Medicine
“AI can support life-or-death decisions made under extreme time constraints,” Dr. Weida said.
AI tools that synthesize risk without replacing clinicians
Dr. Weida outlined several AI models, including a web-based predictive tool developed in collaboration with Kiran Kush, MD, a heart failure cardiologist at Stanford Health Care and incoming ISHLT president. TOPHAT (Tool to Predict Heart Acceptance for Transplant) uses 20 donor characteristics to estimate the probability that a transplant center will accept a donor heart, based on historical data. This tool increases the efficiency of the donor process and reduces the chance that a heart will go unused simply because time runs out before a compatible recipient can be found.
“This tool doesn’t say, ‘This is a good heart,’ or ‘This is a bad heart,'” Dr. Weida explained. “Instead, we quickly show how a donor compares to the national experience. An older donor, or a donor with a single risk factor, such as cocaine use, may appear to be high risk at first glance. But when all variables are considered at once, that donor may not be any higher risk than the typical heart we already use.”
The goal of AI development is to ensure that viable hearts are used and not discarded.
The second tool provides an AI-assisted reading of an echocardiogram, an important test of heart function, as a second opinion for doctors.
“Measuring ejection fraction from echocardiograms is notoriously subjective,” Dr. Wayda said. “We showed that AI-based readings can be more consistent and better match expert interpretations.”
Dr. Weida also explained his future goals. It is an integrated decision support report that combines output from TOPHAT, AI-assisted echocardiogram measurements, other emerging AI tools, and the broader donor medical record into a single, easy-to-understand summary for clinicians making time-sensitive decisions.
“With this kind of comprehensive view, doctors are less likely to make a decision based on a single ‘red flag,’ such as a donor’s age over 50, and refuse to donate a heart that may be in good working order.”
Throughout his talk, Dr. Weida emphasized that AI is a decision support tool, not an autonomous decision maker.
“The real value of AI lies in its ability to quickly and objectively synthesize vast amounts of data to help clinicians make better-informed choices,” he said.
Even a small increase in donor heart utilization could have a significant impact on the transplant waiting list, which includes approximately 4,000 patients.
“Adding 500 hearts and incremental improvement is enough to significantly reduce wait times,” he said.
Technology alone cannot fix policies and incentives
Dr. Weida also argued that changing the way centers are rated and incentivized requires a combination of AI innovation and policy reform.
“I think there’s a temptation to think that this problem can be solved with ‘technology,'” he says. “Transplant policy needs to be restructured with the goal of accessing more donors.”
For AI tools to have a real-world impact, Dr. Weida noted that they also need to be incorporated into existing national transplant infrastructure.
“We can build beautiful web tools, but surgeons aren’t going to log into another site,” he said. “For AI to be useful, it needs to be integrated into standard electronic platforms that are part of the regular data pipelines you already use to vet donors.”
sauce:
International Heart and Lung Transplant Society
