A predictive model for liver disease progression based on logistic regression algorithm

Abstract

Liver disease counts to be one of the most prevalent diseases in the worldwide. Therefore, this paper is  aim to address the problem of predicting liver disease progression. As the existing predictive models focus on predicting the label of disease; the probability of developing the disease is still obscure. This paper, therefore, has proposed a model to predict the probability occurrence of liver diseases. The proposed predictive model used logistic regression abilities to predict the probability of liver disease occurrence. ILPD dataset was used to analyze the performance of the model. The predictive model has shown outstanding performance with a prediction accuracy rate of 72.4%, the sensitivity of 90.3%, the specificity of 78.3 %, Type I Error of 9.7 %, Type II Error of 21.7 %, and ROC of 0.758%. The model has furthermore confirmed the feasibility of the laboratory tests such as as (Age; Direct Bilirubin (DB), Alamine_Aminotransferase (SGPT), Total_Protiens (TP), Albumin (ALB)) to predict the disease progression. The predictive model will be helpful to patients and doctors to realize the progression of the disease and make a suitable timely intervention.