Towards ensemble classification algorithms for breast cancer diagnosis in women: A comparative approach.

Abstract

Despite a spike in growth rate of modern techniques towards breast cancer diagnosis where a perfect diagnostic system would discriminate between benign and malignant findings perfectly, flawless discrimination has not been realized, so radiologists’ decisions are founded on their best judgment of breast cancer risk amidst substantial uncertainty. And in low developed countries where adoption of computer based diagnostics for decision support is low, given the variety of options in the artificial intelligence and machine learning perspective, we endeavored to perform simulations on the breast cancer dataset and 5 classification algorithms that are supported for best performance given small datasets and low computational complexity needs, towards achieving an optimal ensemble model that would nearly perfectly discriminate between cancerous and non-cancerous breast tumors.

Keywords:
Machine learning, breast cancer, artificial intelligence, benign, malignant, supervised learning, ensembles, model optimization, cross-validation.