Ensemble learning for Breast Cancer prediction using Gene expression dataset

Artificial intelligence (AI) has substantially impacted several fields. The rise of deep learning (DL) together with machine learning (ML), plus the availability of enormous DNA datasets, has opened new possibilities for research targeted at applying these models for predicting breast cancer based on genetic data. Breast cancer ranks among the most prevalent and dangerous diseases affecting women, and diagnosing it early can dramatically cut the mortality rates. Due to large genetic databases, deep learning has tremendous promise for predicting breast cancer. However, predicting which genes lead to malignant cells remains tough. Identifying and extracting genes responsible for cancer is critical for accurate cancer prediction. Effective prediction can also facilitate the design and administration of targeted medications. In this work, we retrieved exons from our studies' breast cancer gene sequence. The exonic regions were extracted, used to develop two deep learning models: DNN, and bi-LSTM network. DNA series were translated using K-MER method, while class labels were represented through one-hot encoding. Model performance was evaluated using standard classification metrics. The DNN model achieved a training precision of ninety-eight point five (98.5%) with validation precision of ninety-six percent (96%), whereas the bi-LSTM model obtained a training precision of ninety-four five percent (94.5%) with validation precision of ninety-one percent (91%), indicating the effectiveness of the DNN in this context.