Explainable Artificial Intelligence based Ensemble Machine Learning for Ovarian Cancer Stratification using Electronic Health Records

Abstract

The purpose of this study is to show how ensemble learning-driven machine learning algorithms outperform individual machine learning algorithms at predicting ovarian cancer on a biomarker dataset. Additionally, this study provides model explanations using explainable Artificial Intelligence methods, The method involved gathering and combining 49 risk factors from 349 patients. We hypothesize that ensemble machine learning systems are superior to individual Machine Learning systems in predicting ovarian cancer. The Machine Learning system consists of five individual Machine Learning and five ensemble Machine Learning systems were trained using K-10 cross validation protocols. These training models were then used to predict the development of benign ovarian tumors and ovarian cancer tumors patients. The AUC and Accuracy metrics for ensemble machine learning increased by 19% and 16%. The MCC and Kappa scores for ensemble Machine Learning also increased over individual machine learning by 29% and 33%, respectively. As a result, we draw the conclusion that ensembled-based algorithms outperform individual machine learning in terms of ovarian carcinoma prediction.