Broader Evaluation of TMD Treatment Efficacy and Response (BETTER TMD)

About This Grant

PROJECT SUMMARY/ ABSTRACT (Project 1 BETTER TMD) C-TMD IMPACT Research Project 1 refers to Broader Evaluation of TMD Treatment Efficacy & Response (BETTER-TMD) and will address the misdiagnosis and suboptimal treatment decision gaps in the TMDs. The overall goal of this project is to define and predict which existing treatments are likely to be most effective for specific patients from a whole-person perspective, considering their symptoms and psychosocial context. The proposed study leverages a custom-developed app, MyDocNote, to collect real-world, structured, patientreported treatment outcome data. These outcomes will be paired with de-identified, highly structured data on signs, symptoms, and diagnosis. Our preliminary studies have demonstrated the dinical utility of machine learning (ML) methodology to create a TMD orofacial pain {TMD-OFP) predictive diagnostic algorithm using data collected from 1,020 patients seen at the USC TMD-OFP clinic. We also showed the potential effectiveness of an algorithm to predict various TMD-OFP diagnoses based on clinical signs and symptoms. For this U54 phase, we propose to scale up the pilot work described above and add longitudinal treatment outcome data to our dataset as we seek to develop additional algorithms to predict optimal, individualized treatment decisions. This scale-up will be achieved by expanding our data collection to 19 sites across the US, including 4 university medical centers and 15 clinical private practice sites. All collected signs and symptoms data from these patients will be converted into a structured database format and then made available through the FaceBase Hub with appropriate human subject protections, so we and other interested researchers can conduct further analyses. We project to collect data from up to 1,000 patients per year, for a target total of 5,000 over the 5-year performance period. Our guiding hypothesis is that collecting and analyzing a large, longitudinal, highly structured, whole-person dataset associating clinical signs and symptoms with treatment outcomes will lead to algorithms that can accurately stratify patients and identify the best individualized treatments. To test this hypothesis, three specific aims are proposed. Aim 1 will establish a set of common data elements {CDEs) to characterize the phenotypic traits and treatment outcomes of TMD-OFP patients. Aim 2 will collect a large longitudinal, structured dataset of patient treatment outcomes using our mobile app, MyDocNote. Aim 3 will analyze patient data to improve diagnosis and treatment decision-making for TMDOFP. Successful completion of these Specific Aims will deliver: 1) a large-scale, longitudinal, structured, and whole-person database of TMD-OFP signs, symptoms, and treatment outcomes; 2) decision-making algorithms for diagnosis and treatment based on data analyses including ML methodology. Ultimately, these can form the basis of a real-time point-of-care system to support clinicians in diagnosing TMDs and planning treatment.