An improved percentage rate accuracy in predicting mortality in hepatitis-c using an artificial neural network

Abstract

Background accurate prediction of mortality in Hepatitis-C (Hep C) is essential for policy action and planning. While studies have used artificial intelligent technique (e.g., artificial neural network (ANN)), their appropriateness to predicting mortality in hepatitis-c has been debated. This study presents an improved percentage rate accuracy that is capable of predicting whether a patient suffering from Hepatitis-C Virus (HCV) is likely to survive or die. The constructive research method was adopted for this study, while an Object Oriented Design Approach was adopted for the systems structural design. The Artificial Neural Network system was implemented using java programming language with many program modules and four (4) design classes namely; the Driver class that runs the application program, the Neural Network class, the Neuron and the Layer classes. The network was trained using back propagation machine learning algorithm, a learning rate of 0.8 and a learning error of 0.05. While the weights used for the training were random numbers ranging from -1.0 to +1.0. The maximum number of training sessions was set to 10000 assuming the network does not converge to the leaning error of 0.05. The result of the network showed 85% accuracy in predicting cases of the patients with positive hepatitis C virus that may survive and also 50% accuracy in predicting cases of patients with positive Hepatitis-C Virus (HCV) that may likely to die given the provided data. Neural network is a powerful classification and prediction tools that can help in predicting the outcome of Hepatitis-C virus (HCV) infections. Recommending experiment on the network architecture with a view to either increase the hidden layers or increasing the number of units in the hidden layer. Also, more extensive testing and training should be carried out to achieve the desired result.