Abstract

Breast cancer has recently become the subsequent most dangerous origin of cancer death in women, posing a serious threat to middle-aged women all over the world. However, early detection and prevention can greatly lower the risk of demise. Enhancing the likelihood of cancer repetition is a crucial aspect of the prognosis for breast cancer. For many researchers, predicting the return of breast cancer has been a difficult research topic. Increasingly, data mining techniques have drawn a lot of interest, particularly when they are applied to the creation of prognostic models using survival data. The prior system developed an effective weighted Minkowski radial basis function-based support vector machine and Simulated Annealing Inertia Weight based Chicken Swarm Optimization (SAIWCSO) method for breast cancer recurrence prediction (WMRBF-SVM).Large data sets, however, are not a good fit for the SVM method. Because when target classes coincide and the data set contains additional noise, SVM also performs poorly.The suggested approach created an ensemble learning-based feature selection using Weighted Average Long Short Term Memory Network (WALSTM) for prediction of breast cancer recurrence in order to address this issue. As an input, the Wisconsin Breast Cancer Dataset (WBCD) is used. The input data are preprocessed to eliminate irrelevant and/or missing value data. Using the z-score normalisation method, repetitive entries were removed during this phase. Following that, linear function-based animal migration optimization and ensemble learning-based feature selection are carried out (LFAMO).Classification is carried out utilizing the Weighted Average Long Short-Term Memory Network according to the selected attributes (WALSTM). Utilizing MATLAB, the experiments are reproduced. According to the experimental findings, the suggested system performs better than the current system in terms of accuracy, precision, recall, and specificity, f-measure.

Keywords: Breast cancer, ensemble-based feature selection, Information Gain (IG), Animal Migration Optimization (AMO) and Weighted Average Long Short Term Memory Networks (WALSTMs).