Diagnosis of Parkinson's disease through EEG signals based on artificial neural network and cuckoo search algorithm

Abstract

Parkinson's disease is a degenerative nervous system condition that impairs mobility. If the condition is not detected early enough, it might have permanent effects for the sufferer. A novel approach for identifying Parkinson's disease is provided in this research, which employs machine optimization and learning techniques. The suggested method's diagnosis procedure may be broken down into three primary steps: "preprocessing," "feature extraction," and "classification." Preprocessing the EEG data is the initial stage in the suggested technique. Database samples are treated using discrete wavelet analysis to remove the destructive influence of noise on the input signals using signal analysis for this aim. The suggested method's second phase will employ principal component analysis to remove duplicate features and minimize data dimensionality. The artificial neural network model is trained and the classification model is built using the retrieved features. The effectiveness of the suggested technique is examined in terms of criteria such as accuracy, sensitivity, and specificity during the experimentation phase, and the results are compared to existing learning models. The findings revealed that the suggested technique enhances illness diagnostic accuracy by at least 8.25% and may be utilized as a useful tool in disease diagnosis.