Enhanced Classifier Accuracy in Liver Disease Diagnosis using a Novel Multi Layer Feed Forward Deep Neural Network

Classification techniques are often used for predictingLiver diseases and assist doctors in early detection of liverdiseases. As per studies in the past and our experiments,conventional classification algorithms are found to be lessaccurate in predicting liver diseases. Therefore, there is a need forsophisticated classifiers in this area. For many medicalapplications, including Liver Diseases, Deep Neural Networks(DNNs) are used but the accuracies are not satisfactory. DeepNeural Network training is a time taking procedure, particularlyif the hidden layers and nodes are more. Most of the times it leadsto over fitting and the classifier does not perform well on unseendata samples .We, in this paper, tuned a Multi Layer FeedForward Deep Neural Network (MLFFDNN) by fittingappropriate number of hidden layer and nodes, dropout functionafter each hidden layer to avoid over fitting, loss functions, bias,learning rate and activation functions for more accurate liverdisease predictions. We used a balanced data set containing 882samples. The data is collected from north coastal districts ofAndhra Pradesh hospitals, India. The training process is carriedout for 400 epochs and finally It is .observed that our modelexhibited 98% accuracy at epoch 363 which is more than theperformance of Neural Network models tuned till now bymachine learning researchers and also some regularly usedclassification algorithms like Support Vector Machines (SVM),Naive Bayes (NB), C4.5 Decision Tree, Random Belief Networksand Alternating Decision Trees (ADT)