Breast Cancer Diagnosis using a Hybrid Genetic Algorithm for Feature Selection based on Mutual Information

Feature Selection is the process of selecting a subset of relevant features (i.e. predictors) for use in the construction of predictive models. This paper proposes a hybrid feature selection approach to breast cancer diagnosis which combines a Genetic Algorithm (GA) with Mutual Information (MI) for selecting the best combination of cancer predictors, with maximal discriminative capability. The selected features are then input into a classifier to predict whether a patient has breast cancer. Using a publicly available breast cancer dataset, experiments were performed to evaluate the performance of the Genetic Algorithm based on the Mutual Information approach with two different machine learning classifiers, namely the k-Nearest Neighbor (K-NN), and Support vector machine (SVM), each tuned using different distance measures and kernel functions, respectively. The results revealed that the proposed hybrid approach is highly accurate for predicting breast cancer, and it is very promising for predicting other cancers using clinical data.