Evolutionary algorithms based RBF neural networks for Parkinson´s disease diagnosis

Parkinson´s Disease (PD) is the second most common neurodegenerative action and expected to increase in the next decade with accelerating treatment costs as a consequence. This situation leads us towards the need to develop a Decision Support System for PD. In this paper we propose different methods based on evolutionary algorithms and RBF neural networks for diagnosis of PD. Three different evolutionary algorithms; genetic algoritm, particle swarm optimization and artificial bee colony algorithm (ABC); are used for training different structures of RBF neural networks. The experimental results show that the usage of ABC algorithm based RBF networks results better than the other methods, either in terms of accuracy or speed for PD diagnosis.