A research team at the U of U and Intermountain Primary Children's Hospital has developed a set of mathematical tools using artificial intelligence to assess the risk of heart complications in patients with a number of pre-existing risk factors. Stock photo, Canva.
SALT LAKE CITY — A research team at the University of Utah and Intermountain Primary Children’s Hospital developed a set of mathematical tools using artificial intelligence to assess the risk of heart complications in patients with a number of pre-existing risk factors. The tools are available online now for doctors and patients to use.
The artificial intelligence heart study, published in the online journal PLOS Digital Health, looked at 1.6 million electronic health records from patients at the University of Utah and Primary Childrens’ hospitals. Using a technique called Probabilistic Graphical Models (pictured below), the team created networks that connected the health outcomes with various risk factors leading to heart transplants, fetal heart abnormalities, and others.
Figure contained in article: An explainable artificial intelligence approach for predicting cardiovascular outcomes using electronic health records. Image credit: Wesołowski S, Lemmon G, Hernandez EJ, Henrie A, Miller TA, Weyhrauch D, et al. (2022) An explainable artificial intelligence approach for predicting cardiovascular outcomes using electronic health records. PLOS Digit Health 1(1): e0000004. https://doi.org/10.1371/journal.pdig.0000004
Dr. Martin Tristani, Professor of Pediatric Cardiology at the University of Utah, was among the authors of the study. He said the tools not only predict specific outcomes based on a list of risk factors but also assess how they interact to change the calculation.
“[It] allows us with a single model to ask multiple questions about the risk of getting a particular disease,” Tristani told KSL NewsRadio.
And he said, in some cases, they can help determine when a specific complication might occur.
“We can predict increased risk of cardiovascular complications. Could we really say within a year? Probably not, but certainly, we can arrive at a general time frame,” he said.
Tristani says the same techniques might be used in other ways. For example, to help calculate risks from other serious health conditions such as cancer and diabetes.