AI leverages ECG data to predict irregular heartbeat, death risk
Using electrocardiogram data, Geisinger researchers have trained deep neural networks to predict patients’ risk of developing a potentially dangerous irregular heartbeat or of dying within the next 12 months.
Using electrocardiogram data, Geisinger researchers have trained deep neural networks to predict patients’ risk of developing a potentially dangerous irregular heartbeat or of dying within the next 12 months.
That’s the finding of two preliminary studies to be presented at the American Heart Association’s Scientific Sessions 2019 on November 16 to 18 in Philadelphia, which leveraged more than 2 million ECG results from three decades of Geisinger's archived medical records.
According to researchers, both studies are among the first to use artificial intelligence to predict future events—such as atrial fibrillation (AF)—from an ECG rather than to detect current health problems.
In one study, the neural network predicted that, within the top 1 percent of high-risk patients, one out of every three patents was diagnosed with AF within a year. In addition, the patients predicted to develop AF at 12 months had a 45 percent higher hazard rate in developing AF over 25-year follow-up than the other patients.
“Currently, there are limited methods to identify which patients will develop AF within the next year, which is why, many times, the first sign of AF is a stroke,” says senior author Christopher Haggerty, assistant professor in Geisinger’s Department of Imaging Science and Innovation. “We hope that this model can be used to identify patients with atrial fibrillation very early so they can be treated to prevent stroke.”
In a second study, a deep neural network was able to predict one-year mortality directly from ECG data even when clinically interpreted as normal.
According to researchers, three cardiologists separately reviewed the ECGs—that had first been read as normal—and they were generally unable to recognize the risk patterns that the neural network detected.
“This is the most important finding of this study,” says Brandon Fornwalt, MD, senior author on both studies and associate professor and chair of the Department of Imaging Science and Innovation at Geisinger. “This could completely alter the way we interpret ECGs in the future.”
That’s the finding of two preliminary studies to be presented at the American Heart Association’s Scientific Sessions 2019 on November 16 to 18 in Philadelphia, which leveraged more than 2 million ECG results from three decades of Geisinger's archived medical records.
According to researchers, both studies are among the first to use artificial intelligence to predict future events—such as atrial fibrillation (AF)—from an ECG rather than to detect current health problems.
In one study, the neural network predicted that, within the top 1 percent of high-risk patients, one out of every three patents was diagnosed with AF within a year. In addition, the patients predicted to develop AF at 12 months had a 45 percent higher hazard rate in developing AF over 25-year follow-up than the other patients.
“Currently, there are limited methods to identify which patients will develop AF within the next year, which is why, many times, the first sign of AF is a stroke,” says senior author Christopher Haggerty, assistant professor in Geisinger’s Department of Imaging Science and Innovation. “We hope that this model can be used to identify patients with atrial fibrillation very early so they can be treated to prevent stroke.”
In a second study, a deep neural network was able to predict one-year mortality directly from ECG data even when clinically interpreted as normal.
According to researchers, three cardiologists separately reviewed the ECGs—that had first been read as normal—and they were generally unable to recognize the risk patterns that the neural network detected.
“This is the most important finding of this study,” says Brandon Fornwalt, MD, senior author on both studies and associate professor and chair of the Department of Imaging Science and Innovation at Geisinger. “This could completely alter the way we interpret ECGs in the future.”
More for you
Loading data for hdm_tax_topic #better-outcomes...