Cryotherapy Outcome Prediction with Explainable Machine Learning: A LIME-Based Approach

Abstract

Cutaneous warts, caused by human papillomavirus (HPV), are commonly treated with cryotherapy, though outcome prediction remains difficult due to patient and lesion variability. This study applies four machine learning classifiers-Random Forest (RF), Support Vector Machine (SVM), Multi-Layer Perceptron (MLP), and K-Nearest Neighbors (KNN)-to predict treatment outcomes. Models are trained using Random Search for hyperparameter tuning and evaluated via ten-fold crossvalidation using nine metrics including accuracy, recall, specificity, <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$F_{1}$</tex>-score, MCC, and Cohen's kappa. The TOPSIS method from Multi-Criteria Decision Making (MCDM) identifies RF as the top-performing model. Finally, Local Interpretable ModelAgnostic Explanations (LIME) are used to interpret predictions, highlighting key features influencing cryotherapy success and enhancing transparency.

Ähnliche Arbeiten