Prediction of Chronic Liver Disease Using Machine Learning Algorithms and Interpretation with SHAP Kernels

The study below shows that we have mainly focused on predicting liver disease. The dataset used was obtained from Kaggle which contained 11 columns and 583 rows, amongst which there were 142 females and 441 males. The dataset contained various features which helped our model to do better predictions. We have used Machine Learning algorithms for the purpose which included classification algorithms like Random-Forest, Logistic-Regression, Decision-Tree, SVM, KNN. The results obtained from all these classifiers were then given to an ensemble model to arrive at exceptional results. In the paper below we obtained the highest accuracy using Random Forest classifier which gave us a result of 72%. We have then used LIME & SHAP kernels. SHAP kernels allowed us to explain the underlying decision-making-process in a better way, it also enhanced the interpretability and trustworthiness of the models. These findings highlight the effectiveness of the ML models and their contribution to the field of liver disease diagnoses and prediction.