A study on sleep parameters in Parkinson's Disease patients

Parkinson's Disease (PD) is a neurodegenerative disorder that affects a considerable number of people all over the world. Symptoms are the results of the dopaminergic neurons death in the substantia nigra pars compacta and the dopamine lack leads to the classical parkinsonian movements such as tremor, irregular gait, paralysis and a low muscular strength. Actually the disease is associated also to other symptoms that are not specifically involved in the motor system and can precede the movements disorder. Non-motor features are: olfactory dysfunction, fatigue, pain, psychiatric and sleep disorders; they have an high impact on the quality of life. This study is focused on the analysis of the electroencephalogram signals (EEG) and the inertial signals, both of them collected during the night. Regarding EEG, three algorithms are implemented: one for distinguishing the REM phases from the others, and two for distinguishing all the sleep stages. Features extraction and machine learning techniques are employed, reaching an accuracy of 98.7%, regarding the REM-NREM differentiation, and an accuracy of 93.1% for all the stages recognition. Wavelet transform is used in the last algorithm for all stages identification; an accuracy of 93.5% is obtained. The second part is centred on the comparison between PD patient's and healthy subjects' inertial signals. The effect of PD on sleep is undeniable: the patient performs less and slower movements with respect to controls. For evaluating these differences an activity index is calculated; moreover other parameters are extracted: the number of turning in bed, the angular velocity and the turning duration. Since the EEG data come from an existing on-line database, the future implementation of this work is to associate EEG and inertial data from the same patient, in order to make an evaluation about his quantity of motion in relation to the disease status. Finally, two user interfaces are proposed for assuring an easier employment of this work.