کنفرانس مهندسی برق ایران

صفحه اصلی / سی و سومین کنفرانس بین المللی مهندسی برق

Recurrence Quantification and Machine Learning: A Novel Approach for Parkinson’s Disease Diagnosis from EEG Signals

نویسندگان :

Asghar Zarei¹ Alireza Talesh Jafadideh²

1- دانشگاه صنعتی سهند 2- دانشگاه تهران

کلمات کلیدی :

Parkinson's disease (PD)،Recurrence Quantification Analysis (RQA)،Electroencephalogram (EEG)،Machine Learning (ML)

چکیده :

Parkinson's disease (PD), the second most common neurodegenerative disorder globally, primarily involves deficiency of central nervous system dopamine. Hence, diagnosis of PD presents serious challenges and is usually a prolonged process without a standardized protocol. As a result, various studies have been conducted to find reliable biomarkers for PD. One such approach is through a characterization of EEG signal features. EEG records neuronal activity from electrodes placed on the skull, and with the advent of AI, EEG signal features have been incorporated into machine learning (ML) algorithms for assistance in automatically diagnosing neurological diseases. This suggests that EEG signals can be regarded as important biomarkers that may help discriminate PD patients from controls. In this study, we explore the potential of Recurrence Quantification Analysis (RQA) features calculated from EEG signals as biomarkers for PD. Based on publicly accessible data received from The Patient Repository for EEG Data + Computational Tools (PRED + CT), we analyzed EEG recordings of PD patients who were repeatedly submitted to auditory stimulation. We employed Support Vector Machine (SVM), K-Nearest Neighborhood (KNN), and Random Forest algorithms for the classification procedure and utilized a 10-fold cross-validation method. The proposed model achieved an average accuracy of 95.72 % separating PD patients from healthy controls using the SVM classifier. This indicates that RQA features from the EEG signals could serve as promising biomarkers for PD.