Determining medication dosage and the effect of dosage on motor symptoms based on the severity of motor symptoms in Parkinson's patients using machine learning

Parkinson's disease is a common neurological disorder that results from the loss of cells in the brain that control motor functions and causes motor symptoms such as tremor. Many patients, especially the elderly, experience problems such as excessive medication intake, forgetting to take medication, not taking medication on time and long intervals between routine examinations. In order to prevent these problems, this thesis aims to design a system that enables the optimization of disease management through a technology-based interaction between the doctor and the patient during the medication treatment process. Raspberry Pi based prototype device design, Leap Motion (LM) non-contact sensor based motor symptom measurement, filtering of motor symptom data with Singular Spectrum Analysis (SSA)-Fast Furier Transform (FFT) integration developed as a new method, machine learning supported dosage-motor symptom prediction and applications were made. With the developed prototype drug intake device, the drugs prescribed by the specialist doctor to the patients were successfully administered. The problems of patients missing doses, overdosing and not taking the medication on time were eliminated. The raw motor symptom data received from the patients with LM sensor was filtered with the developed SSA-FFT integration method and a better filtering was provided compared to manual SSA filtering. With Artificial Neural Networks (ANN), dosage-motor symptom prediction was made with a high accuracy rate of 99% using a model with 10 neural