EEG-Based PD Classification Model Coupled with Machine Learning.

One of the most prevalent neurological brain diseases is Parkinson's disease, which can be diagnosed a long time ago with a variety of clinical methods. In recent years, it has been common practice to use Electroencephalography (EEG) signal analysis to identify dementia in its early stages because of its high speed, low cost, and accessibility. Many novel methods which apply EEG to the diagnosis of Parkinson's disease are shown to be simple and effective. Recent years have seen the development of EEG signal processing as a key technique for researchers to gather appropriate features for Parkinson's disease diagnosis. In this study, a novel system was created for computer-aided diagnosis that is capable of extracting features from EEG signals and discriminating patients affected by Parkinson's disease. After per processing the EEG data, the Butterworth filter has been used to decompose the signals into four frequency sub-bands. Welch's PSD features were then extracted as the input of supervised machine learning methods-the k-Nearest Neighbor (KNN) to classify EEG features into Parkinson's disease (PD) and healthy controls (HC). The 10-fold cross-validation has been employed to validate the performance of this model. The results achieve 98.82% accuracy, 99.19% sensitivity, and 91.77% specificity, respectively. The acquired findings demonstrate the validity of our strategy and that our diagnosis method is improved when compared to earlier research. At last, this novel method may be a supplementary tool for the clinical diagnosis of Parkinson's disease.