Augmenting Neurological Movement Disorder Assessment Metrics Using Deep Learning Algorithm-based Predictions and Haptic Interfaces

Abstract

Several neurological movement disorders, such as Essential Tremor (ET), Parkinson’s Disease, and dyskinesia, are characterized by unwanted, chaotic, involuntary tremors that vary in severity and location. The severity of these movement disorders is clinically evaluated with assessments such as the Fahn-Tolosa-Marín clinical rating scale (FTM) and the Essential Tremor Rating Assessment Scale (TETRAS). These manual rating scales, while useful in a clinical setting due to the low cost and ability to be done without any equipment, often suffer from inter- and intra-rater error and still require expert clinical evaluation. This thesis presents methods to augment and further quantify these assessment metrics, allowing for accurate assessment of tremor severity. Towards the development of an assessment available outside of the clinic, a Long Short-Term Memory (LSTM) network was used to predict ET severity with Inertial Measurement Unit (IMU) data collected from twelve participants with ET rated on TETRAS in Chapter 2. The model performs more consistently than naive median guessing but suffers from more variance, and the model's performance improvement over naive median guessing was not statistically significant. In Chapter 3, the spiral drawing task of these assessments was performed on a haptic interface, which modulated the stiffness of the writing surface. Results indicate healthy individuals move more smoothly with traditional high-impedance spiral drawing, while individuals affected by tremors may write more smoothly under a lower-impedance environment. This promotes future investigation into the understanding of movement quality and tremor classification in three-dimensional space. Taken together, these results represent a contribution to a more objective and quantitative diagnostic methodology for assessing tremor severity than what is the gold standard in clinics.