LDP-XAI Dashboard: A Clinical Decision Support System using Extreme Gradient Boosting and Explainable AI

The early diagnosis of liver disease is critical for patient survival, yet standard machine learning diagnostics often function as opaque "black boxes," lacking the transparency required for physicians to trust their predictions.Furthermore, medical datasets are naturally imbalanced, frequently leading to biased algorithms and high false-positive rates.This project addresses these critical flaws by developing the LDP-XAI Dashboard, an integrated and fully transparent Clinical Decision Support System (CDSS). Utilizing the Indian Liver Patient Dataset (ILPD), the methodology employs the Synthetic Minority Over-sampling Technique (SMOTE) to eliminate class bias and ensure fair learning. The core predictive engine is powered by the Extreme Gradient Boosting (XGBoost) algorithm, while SHapley Additive exPlanations (SHAP) are integrated to provide mathematically grounded visual explanations of the model's reasoning. The balanced XGBoost model achieved a robust predictive accuracy of 94.0%, and the SHAP integration yielded a Faithfulness Score of 0.1987. Deployed via an interactive clinical interface, the system translates these metrics into actionable clinical insights. By enabling timely interventions and fully transparent reasoning, this system acts as a catalyst in medical informatics, successfully driving the paradigm shift from problem-solver to problem-predictor in professional clinical environments.