A Comparative Study on Parkinson's Disease Diagnosis Using Neural Networks and Artificial Immune System

In this paper, different types of classification methods are compared for effective diagnosis of Parkinson's diseases. These methods are artificial immune system, feed forward, learning vector quantization and probabilistic neural network algorithms. Total 197 Parkinson's data which consists of 22 features is used in this study. The 10-fold cross-validation technique is performed to compute the performance of the algorithms used for Parkinson's disease diagnosis. The classification accuracy with 95.6%, 95.4%, 91.4% and 96.5% are obtained from artificial immune system, feed forward, learning vector quantization and probabilistic neural networks, respectively. The best result for the classification accuracy is obtained using probabilistic neural network. The results of this paper are also compared with previous studies using same Parkinson's dataset.