Deep Learning Approaches for Breast Cancer Disease Prediction

Abstract

Breast cancer is a serious illness resulting from abnormal growth of breast cells. Early detection is crucial in the medical field to reduce the risk to human life. Previous research utilized the Predictive Modelling Technique (PMT) for this purpose. However, PMT struggles with handling noisy data, which can lead to inaccurate predictions. To address these challenges, the Deep Learning-based Breast Cancer Disease Prediction Framework (DLBCDPF) was developed. In this approach, feature selection is carried out using an optimization algorithm known as the Ranking-based Bee Colony method. The F-score values are used as fitness measures to identify the most relevant features. These features are ranked in descending order of their F-Scores, and a subset is formed. Data clustering is achieved through FCM clustering, while classification is performed using an Improved Deep Neural Network. The entire analysis was conducted on a breast cancer dataset using optimization techniques, and results indicate that this method surpasses previous models in accuracy.