Industry leading news about artificial intelligence in healthcare and diagnostics.

Risk of Developing Cardiac Complications May be Dented by AI

By scouring through 1.6 million patient records, an artificial intelligence (AI) algorithm was able to calculate the risk of developing cardiac complications based on a patient’s unique combination of comorbidities.

The research from the University of Utah and Intermountain Primary Childrens Hospital, both in Salt Lake City, will allow physicians to identify and treat heart diseases before patients may even know they have a risk of developing cardiac complications.

Published in PLOS Digital Health, the new research used a recently developed AI algorithm called Poisson Binomial-Based Comorbidity Discovery (PBC) to find patient-specific risks for cardiac disease. The researchers focused on the risk of needing a future heart transplant or the risk of sinoatrial node dysfunction, which causes irregular heart rates. The team also sought to identify comorbidities that indicated the presence of different types of congenital heart diseases.

Researchers used anonymized data from electronic health records (EHR) from University of Utah and Intermountain Primary Childrens Hospital patients, all of which represented 77 million discrete visits. The vast quantity of data used in this study enabled researchers to identify cardiac risks with a high degree of accuracy.

Algorithm Helped Unearth Heart Transplant Risk Factors

By apply the PBC algorithm to the trove of EHR data, researchers were able to pinpoint several cardiac risk factors that were not previously known. Findings for the risk of heart transplant alone found the following for adult patients:

  • The use of milrinone, a medication used to treat heart failure, was associated with 175 times greater risk of requiring a heart transplant.
  • Patients with a history of cardiomyopathy were at 86 times higher risk of requiring a heart transplant.
  • People with a history of myocarditis caused by a virus had 60 times greater risk of needing a heart transplant.

Such AI-mined factors can help clinicians better identify patients who are at higher risk of developing cardiac complications, allowing caregivers to intervene earlier.


"No matter how aware you are, theres no way to keep all of the knowledge that you need in your head as a medical professional in this day and age to treat patients in the best way possible,” Mark Yandell, PhD, said in a news release. “The computational machines we are developing will help physicians make the best possible patient care decisions, using all of the pertinent information available in our electronic age.”

Yandell is the senior author of the PBC study, a Professor of Human Genetics the University of Utah, and Co-Founder of Backdrop Health in Salt Lake City, a company that develops software to look for comorbidity factors in patient data.

A significant potential benefit of this technology is that it not only identifies risks associated with specific comorbidities, but also provides risk scores that allow for personalized risk assessment and treatments approaches.

“This novel technology demonstrates that we can estimate the risk for medical complications with precision and can even determine medicines that are better for individual patients,” said Josh Bonkowsky, MD, PhD, Director of the Center for Personalized Medicine at Intermountain Primary Childrens Hospital. Bonkowsky was not an author of the study but has an interest in the potential applications that result from it.

Other hospitals are also using AI to better calculate cancer risks. For example, Massachusetts General Hospital in Boston has used AI to analyze mammogram images to identify patients who had higher risk of developing breast cancer.

‘Refine the Risk for Virtually Every Medical Diagnosis’

Further, while PBC offers promising advances in improving the early recognition of cardiac diseases, it has wider implications.

“We can turn to AI to help refine the risk for virtually every medical diagnosis,” explained Martin Tristani-Firouzi, MD, the studys corresponding author and a Pediatric Cardiologist at University of Utah Health and Intermountain Primary Childrens Hospital. “The risk of cancer, the risk of thyroid surgery, the risk of diabetes—any medical term you can imagine.”

The potential to weave AI technologies with EHRs should perk the interest of healthcare leaders who not only have an interest in medical applications of AI, but also want to take further advantage of the vast amounts of patient data that health systems collect.

—Caleb Williams

Related Resources:

An explainable artificial intelligence approach for predicting cardiovascular outcomes using electronic health records

Artificial Intelligence Identifies Individuals at Risk for Heart Disease Complications

Primary Childrens Center for Personalized Medicine

Massachusetts General Hospital Sees Success With AI-Based Precision Medicine Breast Cancer Screening