Abstarct: This Thesis concerns the detection of Parkinson’s using Echo State Neural Networks. Detection and prevention of Parkinson’s disease in the early stages can drastically improve the treatment procedure in coming years. Echo State neural networks demonstrate promising results for early detection due their computational power and ability to learn complex structures. This thesis is used a wide variety of datasets from trusted sources to train a neural network predicting the possibility of having Parkinson baxsed on set of features. Finally, the echo state network is used to detect the level of Parkinson's disease, and linear regression algorithms are used to train these networks. The numerical results show a significant improvement using Bayesian Optimization Algorithm in detection accuracy.