Token-Level Cross-Modal Transformer with Contrastive Multi-Task Learning for Breast Cancer Subtype Classification and Survival Prediction

Short summary

This paper proposes a token-level cross-modal transformer with contrastive multi-task learning for joint breast cancer subtype classification and survival prediction in precision oncology. It addresses three limitations of existing approaches: monolithic feature vectors, simplistic cross-modal fusion, and independent optimization of survival and classification objectives. The abstract describes the motivation and limitations but does not include experimental results in the provided excerpt.

• •Token-level cross-modal transformer enables fine-grained interactions between genomic and clinical modalities
• •Contrastive multi-task learning jointly optimizes classification and survival prediction
• •Addresses limitations of monolithic features, simplistic fusion, and independent objective optimization

Generated with AI, which can make mistakes.