National Cancer Institute funds research on computer-generated patient care summaries

The University of Illinois at Chicago has received nearly $1.5 million in funding from the National Cancer Institute, one of the National Institutes of Health, to expand research and development of computer-generated patient care summaries as a tool for reducing hospital readmission rates.

Led by UIC’s Barbara Di Eugenio and Andrew Boyd, the funding will support the work of a multidisciplinary team including health care, computer science and linguistics experts for four years.

“Our goal is to build an algorithm that can be used alongside any electronic health record to automatically generate customized summaries of hospital stays that can serve as guides for patients’ self-care,” said Di Eugenio, professor of computer of science in the UIC College of Engineering.

“We want these summaries to help patients – initially those with heart failure – understand what happened when they were in the hospital and what they should do when they leave the hospital, so that they don’t experience unnecessary setbacks or end up back in the hospital,” said Boyd, associate professor of biomedical and health information sciences in the UIC College of Applied Health Sciences.

According to Boyd, one of the major factors associated with hospital readmission is patient noncompliance with health care provider recommendations, which can often be attributed to poor education and communication between care advisers and patients. The summaries are “about helping patients understand that the follow-up appointments and medications noted in their discharge summaries can help keep them healthy and out of the hospital,” Boyd said.

Using Natural Language Processing software for text summarization, the researchers have already started extracting the medical concepts from physician discharge notes and nursing care plans that will serve as sources for the easy to read narratives. Their research on provider terminology has been published in the International Journal of Medical Informatics.

Di Eugenio says that new funding will be used in part to expand their analysis of the divergent medical terminology used by different providers caring for patients with heart failure, but it also will support more work studying the terminology and language patterns common for the end-user: patients.

“With the new funding, we can also dig deeper into the patient perspective,” she said. “For our algorithm to really make a difference and affect readmission rates, we need to make sure it allows not only for provider input but that it also compensates for the type of language patients and caregivers expect and understand.”

“Sometimes, the simplest way to get information on patients’ experiences is to ask them,” said Richard Cameron, professor of linguistics in the UIC College of Liberal Arts and Sciences. Cameron will lead motivational interviewing of patients and analysis of the terms and patterns from interview transcripts.

Boyd says incorporating the patient perspective is about bridging the gap between providers’ and patients’ goals, and levels of language complexity.

“When providers speak up to 80 percent of the words they use may be considered complex, but when patients speak that percentage drops to somewhere around 16 percent,” Boyd said. “Similarly, doctors may want patients to ‘reduce sodium levels’ but patients may simply want to ‘get back to normal.’ All these factors need to be reflected in a care summary if it is going to be effective.”

“We think that better-informed patients, who understand the connection between their illness, hospital experience and outpatient instructions, will be more engaged in their disease management at home,” said Di Eugenio, “and that this will lead to decreased readmission rates.”

Also working with Di Eugenio, Boyd and Cameron on this project are UIC’s Karen Dunn Lopez, Pamela Martyn-Nemeth, Carolyn Dickens, Amer Ardati and Debaleena Chattopadhyay.