Topological Data Analysis of Resting-State EEG for On/Off Medication Classification of Parkinson’s Disease Patients

A vast number of individuals globally experience Parkinson’s disease (PD), which is classified as a neurological disorder. Identifying biomarkers that can aid in the early detection and monitoring of PD is crucial for improving patient’s treatment. In this study, we investigated the use of resting-state electroencephalogram(EEG) data for PD classification. We used the resting state EEG(rEEG) data from 15 PD patients on/off medication states. We applied a topological data analysis(TDA) approach to extract different and important features from the rEEG data, specifically entropy. We found that the entropy values was the most decisive feature among all features we examined and result for classification between patients on medication and off medication (with $\mathbf{8 5 \%}$ accuracy). Our innovation centers around the utilization of TDA for examining Parkinson’s disease data with exploring features and structures in higher dimensions and also with extraction of topological features through Persistent Homology.