Multimodal Breast Cancer Prognosis Prediction
Designed a multimodal framework using mRNA expression, copy number alteration, and clinical data from the TCGA-BRCA dataset. Transitioned from MLP-based models to self-attention and cross-attention mechanisms for improved performance. Focused on contrastive learning techniques to handle feature embeddings without classification during training.
Sep 26, 2024
