IHO-CBC: Biomarker Discovery Based on Improved Hybrid Optimization Algorithms for Cancer Classification

Abstract

For the early detection of cancer, highly sensitive and specific biomarkers are needed. Bio-fluid biomarkers are particularly useful because they can be used for non-biopsy tests. Although the altered human activities of cancer cells have been observed in many studies, little is known about cancer biomarkers for cancer screening. Cancer classification with a highly efficient tool is essential in this era. When a gene is selected from microarray datasets, many problems are addressed, like diminishing the number of inappropriate and noisy genes. This paper proposed the Improved Hybrid Optimization in Cancer Biomarker classification (IHO-CBC) Model for cancer biomarker detection. The microarray datasets are pre-processed and training using Feedforward Neural Network (FFNN). The Enhanced Binary Black Hole (E-BBH) Algorithm for best features is selected. The Classification has been done with the Hybrid method E-BBH-FFNN. E-BBHO is a novel optimization algorithm that improves the accuracy of the Classification by identifying the optimal weights and biases. The stability improvement with ensemble methods is particularly noticeable for small signature sizes (a few samples of genes), which is most relevant for designing a diagnosis or prognosis model from a gene signature. The experimental results show that the E-BBH-FFNN outperforms the performance metrics like mean square error (MSE), AUC, and F-Score.