A Magazine for the George Mason University Community

Unlocking Key Information in Health Records

By Mason Spirit contributor on August 6, 2018

Analyzing the wealth of data available in electronic health records is a powerful new weapon driving personalized medicine and helping improve health care delivery. It is the central focus of Mason’s highly specialized health informatics program. The key to predicting a disease, identifying likelihood of a hospital readmission, or determining the best drug to prescribe in a given circumstance may be found in these records.

“One of the main challenges in health informatics is the large amount and complexity of data in electronic health records and other sources that is beyond what is found in other data-driven disciplines,” says Janusz Wojtusiak, section chief for programs in health informatics in the College of Health and Human Services. “We’re looking at ways we can collect, organize, and analyze these data for use at individual and population levels.”

Farrokh Alemi, Professor, and Hua Min, Associate Professor, Department of Health Administration and Policy, College of Health and Human Services. Photo by Ron Aira

In recent studies, the college’s researchers have looked at data to determine which patients will develop diabetes, patient responses to depression medications, and to predict prognosis in several patient populations.

Professor Farrokh Alemi, the principal investigator on a number of these projects, is currently analyzing electronic health records to predict who has or will develop an opioid addiction. His method is already more effective than existing methods and questionnaires.

“Rather than focusing on behavioral health and the stigma addiction carries, we have taken a very medical view of the issue,” Alemi says. “By looking at the physical symptoms related to substance abuse, our goal is to start a conversation between the clinician and the patient in a way that is not confrontational.”

Among other projects, Wojtusiak is working with GPS technology and machine learning methods, which is a way software can become more accurate in its predictions. He hopes to predict wandering patterns of adults with Alzheimer’s by asking the question, “Can we build a system to track these patients that can learn their behavior, predict wandering, and detect progression of the disease?”

Professor Hua Min also uses machine learning methods in her research, focusing on how these differing data systems can “talk” to each other, as well as the integration, extraction, and analysis of electronic health records data.

“We have so much medical knowledge already acquired and represented in the [records] that we want to try to reuse this knowledge and enhance it with machine learning methods,” says Min, who is part of a nationwide research effort that works with small health care practices to apply these methods and improve the quality of their care.

—Danielle Hawkins


No Comments Yet »

Leave a comment