This script focuses on explaining a pre-trained CNN’s predictions on the MedMNIST dataset using Deep SHAP. Shapley values are computed for each pixel, summed to create a single Shapley score per image, and saved alongside labels in CSV files for interpretability in medical image classification.