AI Can Identify Heart and Death Risks
Thu, April 22, 2021

AI Can Identify Heart and Death Risks

Recent studies show AI can identify patients at risk of dying within a year of developing a potentially dangerous type of arrhythmia, or irregular heartbeat / Credits: Peshkova via Shutterstock

 

Artificial intelligence has greatly helped in detecting diseases and helping medical professionals diagnose. This not only saves time but also makes processes a lot more convenient. Recent studies conducted by Geisinger researchers showed how AI could completely alter the way we interpret electrocardiogram (ECG) results in the future. These studies are among the first to use AI to predict future events from ECG results rather than to detect current health problems.

According to Managed Healthcare Executive, an online site that covers unique and award-winning content for clinical, practical, and business information needs, the researchers examined ECG test results in identifying patients at risk of dying within a year of developing a potentially dangerous type of arrhythmia, or irregular heartbeat. They used more than two million ECG test results from archived medical records within the Geisinger system. 

After that, they trained deep neural networks and predicted irregular heart rhythms, known as atrial fibrillation (AF), using the data. “This is exciting and provides more evidence that we are on the verge of a revolution in medicine where computers will be working alongside physicians to improve patient care,” Brandon Fornwalt, MD, Ph.D., co-senior author on both studies and associate professor and chair of the Department of Imaging Science and Innovation at Geisinger, said. 

The researchers analyzed more than 1.1 million ECGs of patients for the first study. They used highly specialized computational hardware and discovered that one out of every three people within the top 1% of high-risk patients was diagnosed with AF within a year. The results also showed that patients had a 45% higher hazard rate in developing fibrillation over the next 25 years of follow-up compared to lower-risk patients. 

Meanwhile, the team analyzed 1.77 million ECGs and other records for the second study to identify patients who are most likely to die of any cause within a year. Through a neural network model, they accurately predicted the risk of death even in patients deemed by a physician to have a normal ECG. “This is the most important finding of this study. This could completely alter the way we interpret ECGs in the future,” Fornwalt, who co-directs Geisinger’s Cardiac Imaging Technology Lab, said.