AAMI News April 2017

Cardiologist Uses ‘Big Data’ to Predict the Future

Data collected from hours of continuous electronic monitoring can be used to predict serious illness and prompt life-saving treatment before symptoms ever appear, based on research conducted at the University of Virginia (UVA) in Charlottesville.

Randall Moorman
Randall Moorman

“As it stands right now, monitoring data are thrown away. Nobody ever looks at them,” said J. Randall Moorman, a clinical cardiologist and UVA professor of internal medicine, physiology, and biomedical engineering. “I am completely committed to this idea that there are illnesses that we can detect early by analyzing the data that we already have and are just throwing away.” At the AAMI 2017 Conference & Expo, Moorman will discuss how he and his colleagues have turned more than 100 terabytes of data into a “risk estimation device” for deadly conditions such as sepsis in premature babies and hemorrhage and acute lung failure in adults. The conference will be held June 9–12 in Austin, TX.

“We set out to develop mathematical methods to detect a specific abnormality in the heart rate pattern of premature infants and devised a score that told doctors and nurses the risk of a diagnosis of sepsis,” Moorman explained. “This is something no one has ever had before—a quantitative estimate of whether things are going to get worse over the next several hours.”

In what he described as the largest randomized clinical trial in neonatology, Moorman found that showing these scores to clinicians decreased the death rate from sepsis by more than 20%.

“The heart rate patterns aren’t apparent to anybody standing and looking at a monitor,” Moorman said. “These are things that play out over long periods of time, and no one would have the patience—or really the ability—to stand at a monitor and see with their eyes what we are able to detect when we put large datasets together and do mathematical analysis of them.”

According to Moorman, it is important for healthcare technology management departments to start considering how to incorporate continuous risk estimation systems into the workflow of their hospitals—both for clinicians and engineers.

“To support these kinds of information systems requires facilities with computer networks and monitor networks and to some degree the computer capacity to run the algorithms,” he said. “Although, generally speaking, it takes a lot of data and a lot of computers to devise an algorithm; it takes very little to actually deploy it in a hospital.”

Moorman will take the stage at the Austin Convention Center on June 10 from 11:00 a.m.–12:00 p.m. To register, visit the AAMI 2017 website.