Two researchers from Stony Brook University are collaborating on a project to use machine learning models to predict patient outcomes, specifically focusing on opioid use disorder and overdose risk. Richard N. Rosenthal, MD, and Fusheng Wang, PhD, are leading the initiative which is backed by a $1.05 million grant from the Patient-Centered Outcomes Research Institute (PCORI).
Wang's research forms the foundation of this effort. His work involves developing models that assess how likely patients are to develop opioid-related issues. By leveraging patient medical records, Wang and Rosenthal aim to create a tool that allows clinicians to anticipate patient risks and tailor treatment plans accordingly.
“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,” Rosenthal explains. This often results in underutilized models due to their complexity and lack of intuitive design for clinical application.
The innovative aspect of their research lies in what they describe as a "stakeholder in the loop approach." This methodology enables clinicians to provide feedback on prediction models, enhancing both accuracy and usability. “I think probably the most important contribution in this type of model is our stakeholder-in-the-loop approach,” Wang adds.
One significant challenge faced by Wang and Rosenthal is managing complex patient data with numerous clinical variables. The stakeholder-in-the-loop method addresses this by incorporating clinical insights directly into model development.
“A doctor wants to know all the information as quickly as possible, as comprehensive as possible,” emphasizes Wang regarding the need for precise summaries generated by machine learning predictions.
The project includes collaboration with various partners including patient representatives, computer scientists, and health departments from New York State Office of Mental Health and Suffolk County Department of Health.
Looking ahead, the team aims to apply their research methods to other diseases like cardiovascular conditions and test their model's effectiveness in real-world settings such as emergency departments.