U of Virginia physicians work with data scientists to develop predictive model for dosing protocols

Mackenzie Garrity - Print  | 

University of Virginia nephrologist and director of dialysis system Brendan Bowman, MD, has teamed up with engineers at the university to develop predictive techniques to improve medication dosing protocols.

A nine-month pilot study found that the UVA dialysis system could improve the treatment of low red blood cell counts while reducing unnecessary usage of expensive medication by 25 percent. In turn, UVA Health System could save between $750,000 to $1 million annually by using predictive techniques.

Over and underdosing is "basically a control problem, which is classical engineering," Don Brown, the director of the Data Sciences Institute at UVA. "We can use historical data to inform how we make decisions now to affect future conditions. The goal is to develop a system that can help clinicians more accurately control and predict their patient's red blood cell counts based on what we glean from large data samples from the past."

Dr. Bowman and his team have been analyzing 3,000 patient records that go back more than a decade. He is using the records to find ideal dosing levels under various conditions.

Because the amount of data is so large, Dr. Bowman is leveraging data analytics techniques. Benjamin Lobo, research scientist at UVA School of Engineering and Applied Science, built a model that can predict treatment outcomes based on past experiences data. This allows physicians to cut back on the overuse of expensive medication.

"We first set out to stabilize the anemia cycling problem for patients in our pilot and we've now shown we can improve that by over 70 percent. The cost savings, however, were a surprise. Now, we hope to replicate this in a march larger patient group," Dr. Bowman said.

More articles about AI:
AI-powered low-dose CT imaging is faster, safer, more accurate than other methods
5 hospitals, health systems already implementing AI
IBM Watson Health, Medtronic app predicts hypoglycemic events with up to 90% accuracy

© Copyright ASC COMMUNICATIONS 2019. Interested in LINKING to or REPRINTING this content? View our policies by clicking here.

To receive the latest hospital and health system business and legal news and analysis from Becker's Hospital Review, sign-up for the free Becker's Hospital Review E-weekly by clicking here.