Traditional deep learning models are often considered “black boxes” due to their lack of interpretability, which limits their therapeutic use despite their success in classification tasks. This study aims to improve the ...