Liver Disease Prediction Using Machine Learning, Deep Learning, and Bayesian Networks

Liver illness is a great health concern worldwide, and the positive patient’s outcome relies on the early diagnosis. The body requires the liver to perform several vital functions. These functions include protein synthesis and digestion, as well as the storage and release of nutrients and wastes. This paper discusses how Bayesian network models, Deep learning, and Machine learning might improve the diagnosis of the liver disease. The Indian Liver Disease Patient Dataset (ILDPD) was used to test multiple models: Decision Tree, Random Forest, Support Vector Classifier, K-Nearest Neighbor, Gaussian NaiveBayes, Gradient Boosting, AdaBoost, LightGBM, Bayesian Network, and ANN. Evaluation relies on accuracy, precision, recall, F1-score, and confusion matrix. ANN with 97.35% accuracy displayed the highest performance within all models in the study, which implies the more sophisticated Machine Learning and Deep Learning models can be the facilitators to expect the outcome of liver diseases. Neural networks, in particular, boost accuracy and cost-effectiveness of diagnosis.