You are here

AAMI 2017: Predictive Analytics Bolsters Clinical Decision Support

At the University of Virginia, Big Data helps doctors in the fight against sepsis.

Despite a continued push by the Centers for Medicare & Medicaid Services for healthcare organizations to achieve the Triple Aim of improved health and better care at lower costs, patients continue to get sick and sent home without full resolution of their medical problems, says Dr. J. Randall Moorman, a professor of medicine, physiology and biomedical engineering at the University of Virginia.

That often results in costly readmissions, or worse. But information in medical devices, when tapped appropriately, can help doctors and nurses take better care of patients by predicting adverse events, Moorman said Saturday during the opening general session at the Association for the Advancement of Medical Instrumentation’s 2017 conference in Austin, Texas.

“This is a new age in which data can be collected and analyzed,” Moorman said. “Big Data is on everyone’s lips.”

SIGN UP: Get more news from the HealthTech newsletter in your inbox every two weeks!

Avert Illness Through Pre-Emptive Care

Moorman and his colleagues have a goal of developing, validating and deploying predictive analytics monitoring solutions for clinical decision support.

They’ve already achieved some success through the use of an internally developed tool called CoMET (continuous monitoring of event trajectories) in monitoring neonatal infants for sepsis, he said, saving one extra life per 48 infants monitored with very low birthweight (less than 1,500 grams), and per 23 infants with extremely low birthweight (less than 1,000 g). However, he wants more.

“The new reality in some places is that a continuous monitor at the bedside shows the clinicians the risk of sepsis in the next 24 hours, a truly predictive kind of information for them,” Moorman said. “They are able to suspect sepsis before clinical signs appear, do tests, administer therapies, and in many cases, avert the illness part entirely.

“If we can do this for infants, why aren’t we doing this for everyone?”

According to Moorman, the tool combines multiple sets of information into a database, which clinicians consult to act pre-emptively, and also uses clusters and large hard drives. The database is roughly 100 terabytes in size.

“It takes a lot to develop a model,” Moorman said. Deployment, however, is a different story.

“The footprint of the software that is running in real time is actually quite small,” he continued, specifying that his colleague can run 250 patients through the system in real time using his laptop.

Continuous Monitoring Saves Lives

Continuous monitoring of multiple data sources — EKG monitoring, vital signs, laboratory tests — has been key to CoMET’s success, said Moorman. Any set of data analyzed from a pair of sources yielded better predictive models than data from a single source, he said, and data pulled from all sources proved the most effective for him and his colleagues.

“If you’re in the business of predictive analytics monitoring and you’re not using continuous-monitoring data, you will never be as good as the outfit that does,” he said.

Read articles from HealthTech's coverage of AAMI 2017 here.

a-image/Thinkstock
Jun 10 2017

Comments