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

Suxing Liu Byungwon Min

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

Machine Learning 2026-06-25 v1 Artificial Intelligence

Authors: Suxing Liu Byungwon Min

Abstract

Integrating heterogeneous genomic and clinical modalities for joint cancer subtype classification and survival prediction remains a key challenge in precision oncology. Existing approaches suffer from three limitations: (1) they treat each modality as a monolithic feature vector, precluding fine-grained token-level interactions across modalities; (2) cross-modal fusion is typically performed through linear weighting or late averaging rather than structured token exchange; and (3) survival and classification objectives are optimized independently, missing a joint regularization signal.

Keywords

breast cancer subtype classification survival prediction cross-modal transformer contrastive learning multi-task learning

Cite

@article{arxiv.2607.16233,
  title  = {Token-Level Cross-Modal Transformer with Contrastive Multi-Task Learning for Breast Cancer Subtype Classification and Survival Prediction},
  author = {Suxing Liu Byungwon Min},
  journal= {arXiv preprint arXiv:2607.16233},
  year   = {2026}
}