Fusion of bidirectional neural networks decision for a high accuracy breast cancer detection based on DNA biomarker

This study investigates the potential of deep learning-based sequence modelling for breast cancer classification using DNA biomarker data. A decision fusion method Bidirectional Long Short-Term Memory and Gated Recurrent Unit is proposed to distinguish between normal and cancer-related protein sequences. A total of 1217 sequences were collected from the Protein Data Bank, pre-processed using multiple alignment of fast Fourier transform and basic local alignment search tool for similarity analysis, and transformed into numerical representations using integer and label encoding. The results demonstrate that the rule-based decision fusion approach has effectively classified DNA biomarkers with an accuracy rate of 99.18%.