Hybrid greylag goose and particle swarm optimization for early detection of Parkinson's disease from speech features.

Parkinson's disease is brought on by a disturbance in the functions of the brain cells that are responsible for the production of dopamine, which is a chemical that enables brain cells to interact with one another. The cells in the brain responsible for the production of dopamine are the ones in charge of the regulation, adaptability, and fluency of movements. When sixty to eighty percent of these cells are gone, there is a lack of sufficient dopamine production, which makes Parkinson's motor symptoms manifest. In this study, a robust diagnosis of Parkinson's disease is presented in terms of optimized machine-learning models. Specifically, a novel hybrid optimization algorithm is proposed for both feature selection and optimization of the parameters of a neural network to boost the classification of Parkinson's disease. The proposed optimization algorithm is a hybrid of the greylag goose and particle swarm optimization algorithm, denoted by GGPSO. This approach is applied to classifying Parinson's disease from speech signals. The proposed feature selection method is compared to recent methods, showing a promising performance with superior results. In addition, the performance of the proposed optimization algorithm is compared to other optimization algorithms to prove its superiority and effectiveness in optimizing the parameters of the neural network. The overall classification accuracy using the proposed approach is 99.4%, which outperforms the other competing methods. On the other hand, statistical tests have been performed to show the stability and statistical difference of the proposed method. The results of these study confirmed the achieved outcomes and showed robust performance of the proposed methodology in classifying Parkinson's disease.