UB nursing researchers earn informatics award

A manuscript on improving patient care through machine learning co-authored by two UB nursing professors and a PhD student has earned the researchers the American Medical Informatics Association’s Harriet H. Werley Award.

“Identifying High-Need Primary Care Patients Using Nursing Knowledge and Machine Learning Methods,” published in Applied Clinical Informatics, was recognized for its innovative application of nursing expertise and machine learning to improve health care delivery. It was co-authored by Sharon Hewner, professor in the School of Nursing; PhD student Erica Smith; and UB alumna Suzanne Sullivan, research associate professor who is also an associate professor in the College of Nursing at Upstate Medical University.

The Werley award, presented by the AMIA’s Nursing Informatics Working Group, honors research that exemplifies the use of informatics to advance nursing science and improve patient outcomes. Named for Harriet H. Werley, a pioneering figure in nursing informatics, the award underscores the critical role of nursing in shaping health care innovation.

The manuscript highlights a novel approach to identifying primary care patients with complex health needs by integrating nursing knowledge with machine learning models. Using data from electronic health records, the UB team developed a predictive framework that incorporates clinical, behavioral and social factors, providing a more comprehensive view of patient risk. This approach addresses limitations of traditional models that often overlook the holistic needs of patients.

Using psychosocial phenotyping, a method that identifies distinct patient profiles based on clinical and social characteristics, the research demonstrates how machine learning can quickly group high-need patients into meaningful categories. Unlike traditional methods that often rely on health care costs to classify patients, this approach focuses on understanding the interplay between chronic health conditions and social needs within the patient’s environment.

The findings provide actionable insights for clinicians to prioritize early interventions, improve care coordination and reduce health care costs.

“Nursing knowledge was critical in the selection and omission of certain variables, in interpreting the machine learning output and in applying the findings to personalize the care planning process,” Hewner says.

The research also highlights the unique contribution of nursing knowledge in developing holistic, patient-centered approaches to care. Considering factors such as social determinants of health, like housing instability and access to support systems, offers a more accurate picture of patient challenges.

The study's findings also align with a growing emphasis on precision health care, where interventions are tailored to the specific needs of diverse patient populations. By leveraging the capabilities of machine learning, the research advances efforts to improve outcomes for high-need patients while addressing inefficiencies in health care systems.

 Hewner says future applications for this work include “leveraging health information exchange to share these findings with primary care practices regionally and to develop new models to improve perinatal and dementia care for persons with social and health needs.”

The research was supported in part by an award from the Agency for Healthcare Research and Quality.