Identifying Phenotypes in Carpometacarpal Osteoarthritis through Pain, Biomechanics, and Machine Learning Analyses

University Of Florida, Gainesville FL

Investigators

Abstract

PROJECT SUMMARY/ABSTRACT Carpometacarpal osteoarthritis (CMC OA) is a leading cause of global disability, causing significant functional limitations and pain, as well as decreased quality of life. Current treatment options fail to adequately address disease impact, offering limited pain relief, only temporary functional stability, and/or debilitating constraints to the physiologic thumb joint. There is a need to develop novel therapies based on the biomechanical and somatosensory factors that influence CMC OA disease progression. CMC OA is a multifaceted disease with heterogeneous symptom presentation. There is limited research appropriately evaluating biomechanics and pain in CMC OA, despite their combined influence on disease impact. Current biomechanics and pain research allows for robust analysis of functional limitations and somatosensory deficits. A comprehensive evaluation of structure, function, and pain in CMC OA is needed to fully understand disease impact and properly address diagnostic and treatment concerns. The proposed research will provide an innovative evaluation of function and pain in CMC OA by (1) determining a direct relationship through movement- evoked pain (MEP) analysis and (2) utilizing advanced machine learning algorithms to identify the complex relationships in biomechanical and somatosensory data, define patient phenotypes, and predict symptoms. Implementing kinematic, kinetic, and muscle activity analysis with MEP will innovatively relate robust biomechanics research with pain for CMC OA (Aim 1). The results from this aim will identify compensatory movement patterns that alleviate pain, thereby informing alternative and effective therapies. Implementing machine learning algorithms to evaluate complex, multifaceted clinical data will allow for the evaluation of heterogeneous symptomology unique to CMC OA (Aim 2 and 3). Utilizing probabilistic cluster analysis and explainable AI (XAI) will increase transparency and confidence in the interpretation of clinical data by identifying the variables driving model prediction. These variables will be used to distinguish patient phenotypes and understand the mechanisms underlying disease progression. The results from these aims will inform current diagnostic standards for CMC OA and lay the groundwork for predictive prognostic models. Overall, the aims proposed in this research work harmoniously to provide a holistic evaluation of CMC OA disease impact. The results of this research will advance scientific knowledge of the factors that influence disease: structure, function, and pain. This F30 fellowship will provide the applicant with an exceptional foundation in biomechanics and advanced data science techniques. The skills developed through the applicant's training plan will prepare her for a career developing personalized rehabilitation strategies as a Physical Medicine and Rehabilitation (PM&R) physician-scientist.