A Stacking Ensemble Approached Mesothelioma Diagnosis Prediction

Pleural Mesothelioma is a cancer that forms in tissue covering the lungs and chest wall. It is most frequently caused by prior asbestos exposure. Millions of people suffer grave repercussions because their conditions are discovered too late. The study advocates the supremacy of machine learning strategies for enhancing the early and precise prediction of mesothelioma diagnosis. This study has used multiple classification models on the Mesothelioma dataset, available in the public domain. The study handles the class imbalance problem using SMOTE, and feature selection uses a Genetic Algorithm. The machine learning strategies used in the current study are Decision Trees, Random Forests, and Support Vector Machines. Further, this study has proposed ensemble models, namely weighted voting classifier and stacking ensemble classifier. All the machine and ensemble learning models have been evaluated using benchmark parameters like accuracy, sensitivity, specificity, F1-Score, MCC score, and AUC-ROC Curves. The experimental results depict that ensemble models performed better in comparison to single classifiers. Also, the best prediction accuracy (95.5<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>) was achieved using the proposed Stacking Ensemble Model. This model proposed in the study can be utilized in clinics as a decision support system.