Leveraging health systems' data to address oral health disparities

Project Summary/Abstract Poor oral health is a national symbol of social inequity, disproportionately affecting vulnerable populations. Despite its profound impact on overall well-being, oral health remains disjointed from medical care delivery. As a result, vulnerable adults often resort to emergency departments (ED) for treatment of non-traumatic dental

conditions (NTDCs), leading to suboptimal care, clinician frustration, increased ED crowding, and high health costs. The purpose of this proposal is to use electronic health record (EHR) data to better identify adults likely to seek treatment in EDs for NTDCs. Aim 1, Leverage multi-health systems’ EHR data to predict ED use for NTDCs,

will use EHR data from multiple hospital systems and novel social determinants of health data (measured at the individual- and community-level) in a series of advanced machine learning (ML) prediction models. This aim will identify the value of medical and social determinants data in predicting oral health needs. Aim 2, Extract oral

health predictors from the clinical notes of medical encounters that occur prior to ED use, will apply natural language processing techniques to create novel data reflective of medical care providers’ documentation of oral health. This aim will create new data for use in prediction modeling. Aim 3, Assess the feature importance within,

and the predictive performance of, ML prediction models built on structured and unstructured EHR data, will test the contributions of unstructured and structured data source in prediction models designed to identify adults at risk of using the ED for NTDCs. Furthermore, the mentorship and rigorous training plan proposed in this career

development award application will enable Dr. Heather Taylor to become an independent investigator capable of using advanced health informatics approaches to enhance the delivery of oral health services, particularly for underserved populations. To achieve her career goals, Dr. Taylor seeks 1) technical training in the application

of machine-learning, including natural language processing, 2) knowledge development in health disparities research, and 3) enhanced professional development and grant writing skills. Dr. Taylor has assembled a diverse, interdisciplinary panel of mentors to foster her proposed career development and research plan. She is

supported by a resource-rich institutional environment with access to one of the nation’s oldest and largest clinical EHR data repositories. This study is innovative because it utilizes data collected outside of traditional dental care settings to provide health systems with actionable information about adults in their community who

need support navigating the oral-medical divide. The anticipated products of this research include the development and validation of risk prediction models that identify patients who experience poor oral health and require targeted support. After completion, Dr. Taylor plans to transition to research independence and submit

an R01 focusing on optimizing risk predictions for ED use related to NTDCs within health systems and connecting identified adults to definitive dental care.