Abstract
Background/Objectives: Exercise-based interventions are strongly recommended for managing temporomandibular disorders (TMDs). However, conventional approaches have limited capacity to address symptoms associated with mandibular kinematic abnormalities and often lack sufficient logical clarity for reproducible clinical applications. Furthermore, although current diagnostic criteria and imaging modalities primarily assess static anatomical conditions, traditional three-dimensional motion analysis is difficult to implement in routine practice. This study aimed to evaluate the effectiveness of a personalized, exercise-based intervention optimized to patients' lateral excursion (LE) characteristics using an artificial intelligence (AI)-assisted motion analysis system. Methods: An AI-based two-dimensional motion analysis platform was used to quantify maximum mouth opening (MMO) and LE in three patients with TMD. Individualized interventions-including massage, stretching, resistance exercises, coordination training, and breathing exercises-were provided over 3 weeks based on each patient's clinical presentation and movement patterns identified through the kinematic analysis. Results: All three patients successfully completed the intervention. Average pain intensity declined across all cases. Mandibular function improved: the mean MMO increased by 38.92% on average, and LE decreased by -1.55 mm on average. Conclusions: This study demonstrates that a personalized, exercise-based intervention guided by AI-assisted mandibular kinematic analysis was associated with reductions in pain and improvements in dynamic mandibular function. This approach provides a logically clear and objective framework that may support physical therapy in TMD management, advancing beyond conventional static assessment methods.