A Novel Hybrid Stacked Ensemble Model for Breast Cancer Classification

Abstract

The healthcare industry also leverages technology to address various problems. The medical industry continues to adopt artificial intelligence, deep learning, Machine Learning (ML), and big data solutions to automate tasks, improve workflows, and enhance decision-making. Currently, various AI−based solutions are available in the healthcare industry, for example, the analysis of medical images and the identification of patterns in patient data. Thus, these emerging techniques provides many solutions, such as predictive healthcare, and automated drug discovery. The present study proposes the earlier cancer−detection and prediction model to resolve this real-life problem. This study proposed an improved ML model using Synthetic Minority Oversampling Technique (SMOTE) to detect and predict breast cancer at an earlier stage. The four machine learning algorithms achieved the highest test-set accuracy. These algorithms include a novel Hybrid Stacked Ensemble Model (HSEM) and Random Forest (RF), achieving accuracies of 99.12% and 98.59%, respectively; logistic regression, achieving 98.59%; and the support vector classifier, achieving 98.25%. The Area under curve (AUC) for the Breast Cancer (BC) dataset with the HSEM and RF classifier is 99.90%, indicating the model's accuracy. Secondly, cancer treatment exists in expensive types of treatments, and cost has an important role. Therefore, a low-cost solution is required and would be beneficial for the healthcare industry. Thus, this paper developed a novel, low-cost model for cancer prediction for the healthcare industry, enabling people to estimate their cancer risk earlier