Informatics Publishing Limited

R. Bhuvana ¹, S. Purushothaman ², R. Rajeswari ³

Affiliations
1 Department of Computer Science, Vels University, Chennai, IN
2 PET Engineering College, Vallioor-627117, IN
3 Mother Teresa Womens University, Kodaikanal, IN

Abstract

Depression is a serious and widespread public health challenge. This paper propose neural network algorithm for faster learning of psychological depression data. Implementation of neural networks methods for depression data mining using back propagation algorithm (BPA) and Echo state neural network (ESNN) are presented. Experimental data were collected with 21 input variables and one output for working with artificial neural network (ANN). Using the data collected, the training patterns and test patterns are obtained. The input patterns are pre-processed and presented to the input layer of ANN In order to find the optimum number of nodes required in the hidden layer of an ANN, a method has been proposed, based on the change in the mean squared error dynamically, during the successive sets of iterations. The output of BPA is given as input to ESNN. The network trained with transformed vectors is seen to require the least computational effort. The work proves to be an efficient system for diagnosis of depression.

Keywords

Hamilton Rating Scale (HRS) Depression Data, BPA, ESNN.

Full Text

     

R. Bhuvana ¹, S. Purushothaman ², P. Rajeswari ³

Affiliations
1 Department of Computer Science, A. M. Jain College, Chennai - 600 114, Tamil Nadu, IN
2 Institute of Technology, Haramaya University, ET
3 Department of ECE, Institute of Technology, Haramaya University, ET

Abstract

Depression is common in working personality due to high tension in the working environment. Symptoms can affect dayto- day life and can become very worrying. With true depression, there is a low mood and other symptoms each day for more than two weeks. Symptoms can also become rigorous enough to interfere with normal day-to-day activities. This paper suggests an Artificial Neural Network (ANN) algorithm approach for quicker learning of psychological depression data. Performance of neural network methods for estimating depression state with Echo State Neural Network (ESNN) is presented. Tentative data were collected from the patients with 21 input variables. One target output is used for training the ESNN. The training and testing patterns are made using the data as per Hamilton Rating Scale.The input patterns are pre-processed and presented to the input layer of ANN. The proposed method proves to be a capable system for diagnosis of depression.

Keywords

Depression Data, ESNN, Hamilton Rating Scale

Full Text