Machine learning uses imaging to predict heart damage in COVID-19 patients

Scientists at Johns Hopkins have scored a $195,000 grant to help them develop a machine learning tool that can predict worsening heart health in COVID-19 patients.

Funding comes by way of the National Science Foundation, with Hopkins now beginning the first phase of the one-year research project. They plan to collect CT and echocardiography imaging data from 300 patients, along with numerous other pieces of info, to help clinicians determine which COVID patients are at the greatest risk of cardiac injury, the institution announced Monday.

"As a clinician, major knowledge gaps exist in the ideal approach to risk stratify COVID-19 patients for new heart problems that are common and may be life-threatening. These patients have varying clinical presentations and a very unpredictable hospital course," Allison Hays, an associate professor of medicine and clinical collaborator on the project, said in a statement.

Hopkins highlighted the “increasing evidence” of the disease’s negative impact on the cardiovascular system. Along with imaging data, its scientists also plan to collect ECG, lab tests, and continuous vital signs to help train their algorithm.

They then hope to test their clinical tool by using data from COVID-19 patients who have suffered heart injuries and were treated at Hopkins, other nearby hospitals, or possibly institutions in New York. The end game is to offer a predictive risk score that determines a day in advance which individuals are in danger of developing sustained abnormal heart beats, cardiogenic shock or myocardial infarction, or other issues stemming from the virus.

Johns Hopkins believes its researchers are the first to develop an algorithm that would forecast COVID-related cardiovascular outcomes, according to the announcement. Once their work is finished, they plan to make the algorithm widely available to those interested in using it at their own institutions.