An Improved Deep Learning Algorithm for Breast Cancer Survival Prediction Based on Multi-Omics Data.
Breast cancer is a leading cause of mortality among women worldwide. Accurate survival prediction can improve clinical decision-making and support personalized treatment planning. This study aims to develop an interpretable and effective deep learning model for breast cancer survival prediction using multi-omics data.
This study proposes a novel deep learning model combining Bi-directional Long Short-Term Memory (BiLSTM) and Convolutional Neural Network (CNN) architectures, integrated with Minimum Redundancy Maximum Relevance (MRMR) feature selection. The model was evaluated on two large datasets: METABRIC (n=1980) and TCGA-BRCA (n=1080), using clinical, copy number alteration (CNA), and gene expression data. Performance was assessed through metrics such as AUC-ROC and accuracy.
The proposed model demonstrated superior performance compared to existing algorithms, achieving high AUC-ROC and accuracy values across all data modalities. The integration of BiLSTM and CNN architectures allowed the model to capture temporal and spatial patterns, improving prediction robustness. Notably, the model achieved an accuracy of 98% on the METABRIC dataset and 96% on the TCGA dataset.
The combination of BiLSTM, CNN, and MRMR offers an interpretable and accurate framework for breast cancer survival prediction using multi-omics data. This approach provides actionable insights for clinicians and highlights its potential for broader applications in oncology.
This study proposes a novel deep learning model combining Bi-directional Long Short-Term Memory (BiLSTM) and Convolutional Neural Network (CNN) architectures, integrated with Minimum Redundancy Maximum Relevance (MRMR) feature selection. The model was evaluated on two large datasets: METABRIC (n=1980) and TCGA-BRCA (n=1080), using clinical, copy number alteration (CNA), and gene expression data. Performance was assessed through metrics such as AUC-ROC and accuracy.
The proposed model demonstrated superior performance compared to existing algorithms, achieving high AUC-ROC and accuracy values across all data modalities. The integration of BiLSTM and CNN architectures allowed the model to capture temporal and spatial patterns, improving prediction robustness. Notably, the model achieved an accuracy of 98% on the METABRIC dataset and 96% on the TCGA dataset.
The combination of BiLSTM, CNN, and MRMR offers an interpretable and accurate framework for breast cancer survival prediction using multi-omics data. This approach provides actionable insights for clinicians and highlights its potential for broader applications in oncology.
Authors
Nasarudin Nasarudin, Al-Jasmi Al-Jasmi, Abdul Aziz Abdul Aziz, Ab Aziz Ab Aziz, Khan Khan, Hendrawan Hendrawan, Al Riza Al Riza, Manzoor Manzoor, Ali Ali, Mohamad Mohamad
View on Pubmed