Wavelet-Enhanced EEG Feature Optimization Framework for Early-Stage Recognition of Parkinson’s Disease

Suggested methodology focuses on exploiting electroencephalogram (EEG) signals, which offer direct insight into brain activity, to perceive Parkinson's disease (PD) early. Electroencephalogram provides real-time, high-resolution data that are perfect for classifying neurological disorders, in contrast to gene, speech, or handwriting-based modalities that require considerable feature engineering and lengthier processing periods. This study presents a wavelet-enhanced system that uses efficient wavelet transforms to decompose electroencephalogram signals before extracting statistical features. The Satin Bowerbird Optimization (SBO) technique is used to lower the dimensionality of the features and choose the best wavelet basis. Using their shown effectiveness in medical diagnostics, a group of deep learning models InceptionV3, DenseNet, MobileNet, Xception, and NasNet are used for classification. The complete system, which was created in MATLAB R2021a, shows better accuracy (97%) than both the Automated Tunable Q-wavelet Transform approach and conventional CNN and machine learning models (88%–92%). These outcomes confirm the effectiveness and resilience of the suggested wavelet-enhanced EEG feature optimization method for precise and timely Parkinson's disease identification.