Parkinson Disease Detection using EEG Signals

Parkinson’s disease (PD) is a prevalent neurodegen-eration disease that affects the nervous system’s motor functionand may lead to paralysis in addition to a decrease in dopaminergic neurons in the substantia nigra of the nervous system. As aresult of PD, structural, behavioral, and neurological alterations occur. However, these variations are modest in the early stages of the illness, making an accurate diagnosis challenging. Early identification of (PD) by electroencephalogram (EEG) signals is critical for effective therapy since the condition impairs one’s quality of life. This study introduces the Residual Dilated MultiScale Feature Network (RDMNet), a novel deep learning architecture specifically designed for early detection of Parkinson’s disease using neuroimaging data from the SanDiego and Iowa datasets. The proposed model, RDMNet, exhibits remarkable performance, boasting and impressive average accuracies of 99.82%, 99.802%, with corresponding F1 scores of 100% on both datasets, underscoring its robustness and generalizability. Furthermore, the results demonstrated by the proposed model outperform 12 contemporary methods, thereby establishing it as the most effective model designed to date. The novel proposed model exhibits remarkable analytical performance in various advanced metrics such as brier score, KL divergence score etc. thereby exhibiting its potential for strong discriminative performance for the detection of the disease, thereby providing valuable insights into the complex neurobiological realm and aid in better treatments in the future.