A Novel Machine Learning Approach Using Boosting Algorithm for Liver Disease Classification

Machine learning has been highly recommended in medical sector for diagnosis of several diseases and for effective decision making due to its performance. With recent years, there is rise in number of liver patients with different kinds of complication and it is important to identify it at initial stage to reduce the risk incurred by it. In this research, we implemented gradient boosting based machine learning classifier to achieve the results. CatBoost and LightGBM model are employed for prediction and classification of liver disease with feature selection approach. Preprocessing is performed on the original dataset to remove deviated values using isolation forest and to get relevant features for better results. Model performance is calculated in respect of precision, accuracy, recall and f1-score. CatBoost resulted in highest accuracy of 86.8% and LightGBM achieves 82.6% accuracy with feature selection on Indian Liver Patient Dataset.