Quantification of symptoms of movement disorders employing motion sensors

Clinical evaluation plays a fundamental role in the diagnosis and evaluation of a movement disorder. This evaluation depends, at least to some extent, on a human decision process. To homogenize and improve the reliability of the evaluation rating scales have been developed through the years. In spite of the improvements, the evaluation of symptoms of movement disorders using rating scales, like every human-decision-making activity, is influenced by the expertise, experience, attention and skillfulness of the evaluator.

The continuous effort to improve the scales reveals the necessity of reliable objective assessments that do not depend on the evaluator. In this thesis we employed movement sensors, sensor fusion, signal processing and machine learning to analyze characteristic symptoms of movement disorders. The goal of each chapter was to evaluate a specific method that could overcome some of the disadvantages that are present in current evaluations. In Chapter 2 we proposed a method to automatically assess tremor from accelerometry signals. In Chapters 3 and 4 we employed IMUs to study the objective assessment of bradykinesia in patients with Parkinson’s disease. Finally, in Chapters 5 and 6 we studied the automatic identification of early-onset ataxia and developmental coordination disorder in young children using inertial measurement units (IMUs), first analyzing coordination during the finger-to-nose test (Chapter 5) and then during gait (Chapter 6).

The results obtained in this thesis show that the quantification of symptoms of movement disorders employing motion sensors can be used to support the diagnosis and monitoring of movement disorders.