The intersection of AI, clinical practice and health outcomes with linked Medical-Dental electronic records. - Abstract Oral health disparities in the United States are a significant public health issue, particularly among underserved populations, including Hispanics, non-Hispanic Blacks (NHB), and rural non-Hispanic Whites (NHW). Poor oral health not only leads to oral diseases but is also closely linked to systemic conditions such as diabetes, cardiovascular disease, and oral cancer (OC). Hispanics and NHBs are twice as likely to be diagnosed with late- stage OC, resulting in survival rates below 40%, compared to over 80% for non-Hispanic Whites. This highlights the urgent need for integrated healthcare that connects oral and general health to improve outcomes for these vulnerable groups. Despite efforts to integrate medical and dental care, oral healthcare remains largely disconnected from the broader healthcare system. The proposed study aims to connect oral health with systematic health to better integrate medical and dental care by developing linked databases, creating real-time clinical tools to improve oral clinical care, and identifying individuals at elevated risk for OC using fair and responsible artificial intelligence (AI) methods. We will leverage existing electronic health records (EHR) and electronic dental records (EDR) from three major dental institutes [Temple University School of Dentistry (TU), University of Colorado School of Dentistry (UC), and West Virginia University School of Dentistry (WVU)] that serve underserved populations to develop and implement clinical tools that may facilitate dental practice and reduce oral health disparities related to OC. The four aims are: (1) develop real-world AI-ready datasets to understand and reduce oral health disparities; (2) create a clinical tool that provides real-time medical history, lab values, and medication information to dental providers; (3) predict OC risk using federated learning across multiple institutions with responsible and interpretable AI methods; and (4) explore the feasibility of implementing the clinical tool and AI prediction model at Temple University. This multi-site initiative addresses a critical gap in healthcare delivery for underserved populations by integrating medical & dental care. It has the potential to reduce oral health disparities by ensuring early OC diagnosis and access to up-to-date patient information from dental providers. The clinical tools and OC prediction risk tailored to diverse populations have the potential to detect early OC diagnosis and treatment, leading to improved OC prognosis and OC outcomes aligning with NIDCR's strategic priorities and setting a new standard for reducing health disparities and advancing precision dental care.
