Two researchers at Stony Brook University are working on using machine learning models to predict patient outcomes. Richard N. Rosenthal, MD, a professor in the Department of Psychiatry and Behavioral Health in the Renaissance School of Medicine, and Fusheng Wang, PhD, a professor in the departments of Biomedical Informatics and Computer Science, are collaborating on this project. Their research focuses on optimizing predictions related to opioid use disorder and overdose risks.
The study is funded by a $1.05 million grant from the Patient-Centered Outcomes Research Institute (PCORI), which supports patient-centered comparative clinical effectiveness research in the U.S. The basis for the award was Wang's research on using machine learning to predict patient risk. The goal is to develop models that can assess how likely patients are to develop opioid use disorder or experience an overdose.
Rosenthal highlighted a common issue with AI model development in healthcare: "Most AI model development in health care is done by the developers so that there is little if any feedback into the process by the end users, such as clinicians," he said. He noted that their work employs a “stakeholder-in-the-loop approach,” allowing clinicians to provide feedback to improve model accuracy and usability.
Wang emphasized the importance of involving stakeholders throughout the model's design and evaluation: “I think probably the most important contribution in this type of model is our stakeholder-in-the-loop approach,” he stated. This collaboration includes input from clinicians, patients, computer scientists, researchers, and community representatives from New York State Office of Mental Health and Suffolk County Department of Health.
The team acknowledges challenges due to complex patient data with numerous clinical variables. Wang stressed the need for precise summaries when predicting patient risk: “A doctor wants to know all the information as quickly as possible, as comprehensive as possible,” he said.
Future plans include expanding this method for other diseases like cardiovascular conditions and testing its effectiveness in clinical settings such as emergency departments.