Big data boom: UNC Greensboro is developing smarter solutions to real-world problems.

• Position: RESEARCH: NORTH CAROLINA: UNC GREENSBORO

If we must be hospitalized, most of us hope to leave healthier than when we entered--or at least well on the road to recovery.

But some patients head home and grow sicker instead of better, requiring more treatment. Some even end up back in the hospital.

What's the difference between patients who make a full recovery and patients who don't? One major element is what clinicians call "frailty"--a constellation of factors that include age, nutrition, psychological health, social supports and more.

When UNC Greensboro Assistant Professor of Nursing Deborah Lekan did her dissertation on frailty several years ago, she conducted a painstaking analysis of information drawn from electronic health records, which had just begun to change how providers cared for patients.

"I basically had PDF copies of nursing documentation and physician notes," Lekan says. Getting the information was time consuming and limited by how many records she could analyze herself.

But now, she and collaborators at UNCG and Cone Health System are harnessing the power of computers, sophisticated statistical techniques and machine learning to dive much deeper.

Their goal? Identify patients at risk of not fully recovering, in real time, and improve the care provided to them.

The project sets machine learning algorithms loose on clinical measurements, notes from nurses and doctors, demographic information and other data to see if they can predict which patients will need additional care.

"One of the benefits of our models is they actually tell us how important different variables are," Assistant Professor of Computer Science Somya Mohanty says. Clinicians should be able to see key features putting a person at high risk of readmission and strategically target interventions.

Patient readmission, especially among older patients, is a critical and costiy issue. With changes to Medicare, hospitals can even face penalties if patients must be readmitted within a month. But the frailty project will break ground in other ways, too. One outcome is essentially a road map which illustrates howUNCG and Cone Health collaborated to tap into the massive amounts of data in Cone's electronic health record system. It provides a guide to legal and technical issues other researchers and hospitals will face in using such data.

The software Cone uses is one of the most commonly used electronic health record systems, so the research could have applications at thousands of hospitals.

Lekan eventually...