Early Prediction of Parkinson’s Disease using Artificial Neural Network

Objectives: The objective of this work is to present an in-depth understanding of the diagnosis of Parkinson’s Disease (PD) is critical for efficacious neuroprotection in the early stage. Diagnostic tools based on machine learning techniques using Striatal Binding Ratio (SBR) of Caudate and Putamen (left and right) are very useful to identify early PD. Methods: This paper presents an approach to develop an ANN model for prediction of Gamma-Amino Butyric Acid (GABA) concentration level for PD and Healthy Group (HG). Using multilayer perception network having 4-30-1 architecture for predicting GABA concentration level. The network is trained to an optimum level and trained network that predicts the GABA concentration level for the interpolated values of input parameters like Striatal Binding Ratio (SBR) of Caudate left, Caudate right and Putamen left, Putamen right. Findings: According to the ANN model, the prediction performance is highly encouraging with minimum error and high accuracy. The intended prediction model for GABA concentration level overcomes misdiagnosis of early PD. Applications: We propose to study the improvement of the early prediction of Parkinson’s disease, by implementing the ANN. The predictive model for diagnosing Parkinson disease using artificial neural network is efficient in early detection of neurogenerative disorders.