Implementation of Neutrosophic-Based Decision Support System for Effective Diagnosis of Liver Disease

Liver diseases have been shown to be highly correlated to excessive consumption of alcohol and other harmful or injurious substances such as drugs and toxins. The Nigeria social milieu cannot do away with excessive consumption of alcoholic-related substances and drugs, which are predominantly consumed on weekends either in parties or clubs. Most Nigeria teenagers and adults alike who are supposedly considered as socially correct indulge in excessive consumption of alcohol and other harmful substances leaving the alcoholic companies and shops smiling to the banks. This unpalatable trend has dire consequences as it raises the figure of liver disease patients, which is mostly confused with other tropical diseases like malaria and as such its manifestation cannot be predicted on time with certainty. Timely diagnosis is a panacea to the management of the disease but this is not the case most times as there are handful of hepatologists that can adequately diagnose this disease since General practitioner might not be able to diagnose them on time. This research seeks to comparatively analyze the performance of Neutrosophic-based Decision Support System and Multilayer Neural Network (Traditional Neural Network) in the classification of Indian Liver Patient Dataset (ILDP) as well as articulate its suitability in the classification or diagnosis of liver disease (ILPD). Object Oriented Analysis and Design methodology was used while the implementation was done using WEKA and Java on a Netbeans platform. Experimental results show that Neutrosophic-Based Decision Support System (NBDSS) with an accuracy of 96.41% using a confusability measurement threshold of 0.003278 performed better than the conventional neural network with an accuracy of 72.45%. This clearly shows that Neutrosophic-based Decision Support System is suitable for the diagnosis of liver diseases.