An Advanced Voting Ensemble Model with GWO Feature Selection for Improved Breast Cancer Diagnosis

This work introduces a voting ensemble of AdaBoost, CatBoost, and XGBoost classifiers with feature selection using Grey Wolf Optimization (GWO) to improve breast cancer diagnosis. The model employs such modern concepts in artificial neural networks to outperform conventional and single-algorithm methods in breast cancer prediction. The model trained with the medical and demographic data identified by EHRs had even higher accuracy rate of 97.90% compared to earlier models which ranged from 74.14%, to 86.6%. Due to the involvement of GWO in feature selection, the most relevant features were found to construct the model, improving its performance and enhancing the interpretability of the model as well. Due to the high model accuracy and application speed, there is a prospect of its utilization in clinical work. The results of the research create a possibility to enhance an overall number of calculations with the help of several progressive algorithms and thus, to increase the efficiency of examination and possible treatments of breast cancer. As suggested in the present study, future research in developing and validating the translation process of the model will require independent, bigger, and less homogeneous samples to test its generalizability for clinical practice.