OCR: OmniNet-Fusion: A Hybrid Attention-Based CNN-RNN Model for Multi-Omics Integration in Precision Cancer Drug Response Prediction

The increasing complexity of cancer treatment necessitates advanced computational models for accurate drug response prediction. OmniNet-Fusion (OCR) is a hybrid deep learning model designed to integrate multi-omics data—genomics, transcriptomics, proteomics, and metabolomics—enhancing precision medicine. The model leverages a Convolutional Neural Network (CNN) to analyze spatial omics data and a Recurrent Neural Network (RNN) to process sequential data, with an attention mechanism highlighting crucial features across omics layers. To optimize predictive accuracy, feature selection techniques such as Lasso regression and mutual information filtering are utilized, while Principal Component Analysis (PCA) reduces dimensionality, ensuring computational efficiency. The model undergoes evaluation using key performance metrics, including accuracy, precision, recall, F1-score, and AUC-ROC, demonstrating superior predictive performance over existing methods. By integrating multi-omics fusion and deep learning, OCR enhances biological interpretability and facilitates personalized cancer treatment. This approach not only improves drug response prediction but also provides deeper insights into cancer mechanisms, supporting precision oncology and advancing AI-driven cancer therapy.