An ANN based Classification Algorithm for Swine Flu Diagnosis

Abstract

Machine learning technology adds a new potential to medical diagnosis systems. This paper presents an Artificial Neural Network (ANN) based swine flu diagnosis model. The proposed model selects significant features for swine flu diagnosis by a feature selection algorithm using k- Nearest Neighbour (k-NN) classifier, which reduces the size of data to be used for training the ANN model with an objective of making the training more efficient and accurate. A threshold value is determined by ANN to identify positive and negative cases and the model classifies the test cases either positive or negative based on the threshold value. The results obtained with the proposed model demonstrate the ability of the model to provide high level of accuracy for swine flu diagnosis. The assessment (classification) ability of the proposed ANN based model is compared with that of Case Based Reasoning (CBR) approaches and is observed that the proposed model is superior to others.

Keywords: Artificial Neural Network, Back-propagation, Pattern, Pattern Classification, CBR

View PDF