“Liver Disease Risk Prediction Using Regularized Deep Learning: A Novel Approach”

Liver diseases are common, thereby posing a significant quality-of-life impairment. They have required accurate and prompt assessments to be managed. Typically, conventional methods of evaluation often include subjectivity and laborious amounts of work. It introduces a DNN-based model with L2 regularization to predict risk categorization of liver diseases based on five categories of categorizations: Healthy_Liver, Low_Risk, Moderate_Risk, High_Risk, and Severe_Liver_Disease. The model includes the hidden layers with rectified linear unit (ReLU) activation functions, batch normalization, and dropout for regularizing to achieve robust learning. The DNN was trained in a very comprehensive Kaggle dataset with many physiological and biochemical features and has achieved an impressive 94.07% classification accuracy. Healthcare professionals will be equipped with a significant resource to develop personalized treatment strategies facilitated by this investigation, which highlights the capabilities of deep learning methodologies in the automation of liver disease assessment. Subsequent research will focus on integrating additional data sources to improve the model’s scalability and predictive precision, thus promoting the advancement of automated liver disease evaluation in clinical settings.