International Journal of Scientific Research and Engineering Development

Liver illness is among the major health problems throughout the world and is often detected only in later stages, which makes treatment difficult. Early identification is important, However, conventional approaches have drawbacks as the symptoms are not always clear. In this project, a machine learningbased approach is used to predict liver disease using the XGBoost algorithm. The dataset contains demographic and biochemical details related to liver function. However, conventional approaches have drawbacks using a hybrid k-Nearest Neighbour (KNN) imputation approach, and the data was normalized before training. The model's performance was evaluated using accuracy, sensitivity, specificity, and confusion matrix. The XGBoost model gave an accuracy of 92.4%, which is better compared to Logistic Regression. To make the model more interpretable, SHAP values were used to determine the most important features influencing the prediction. Finally, the prototype was deployed in a Django web application with features like secure login, prediction form, history management, and storage of user data. This experiment demonstrates how machine learning combined with web technology can provide a helpful instrument for anticipating prediction of liver disease.