Hybrid Ensemble Machine Learning Approach for Cancer Prediction

Abstract

Cancer has been one of the major health challenges worldwide in recent times, with millions of new cases and even fatalities recorded annually. Several studies were conducted previously to detect cancer using various machine learning. Here, the hybrid ensemble model has not been extensively considered. As a result, this work constructed a hybrid ensemble model by combining multiple individual models such as random forest, gradient boost, and logistics regression also known as base learners or weak learners, to create a more powerful and robust model known as the hybrid ensemble model. The foremost objective of ensemble model design is to leverage the diversity and complementary strengths of base learners to improve overall predictive performance. The study revealed that hybrid ensemble machine-learning models consistently outperformed single models in terms of prediction accuracy and precision. The proposed ensemble model achieved a sensitivity, specificity, Area Under the Curve (AUC), precision, F1-score 0.92, 1.0, 0.98, 1.0, 0.98 respectively, and accuracy of approximately 0.97. To further check the stability of the model, we carried out a cross-validation, and an average accuracy of 96.072% was obtained. The proposed hybrid ensemble model will help predict cancer patients’ to save lives altogether and preclude being taken for granted.