Artificial neural network for classification of depressive and normal in EEG

Depression is one of the greatest problems nowadays which might lead to high rates of other negative health outcomes such as obesity, heart disease, and stroke. However, diagnosis of the depression is still greatly depended on the questionnaires. Hence, this project is aimed to implement the Artificial Neural Network (ANN) method to outcome a software based method to determine a subject's depression condition. EEG device was used to measure the brain waves of the subjects. The raw EEG data was used on Neural Network for training through the implementation of pattern classification network where the input of the network and the outputs of the network are depressive and non-depressive categories. From the result, it is found that 10 hidden layers Scaled Conjugate Gradient algorithm (trainscg) with inputs data from electrode C3 and C4, providing the result with 95% efficiency where out of 20 tester samples, 19 were detected correctly by the algorithm.