A cancer-type-aware framework for robust multimodal survival prediction under missing modalities
Briefings in Bioinformatics · 01.03.2026 · 27 (2)
DOI: 10.1093/bib/bbag124 ↗Kurz gefasst
Fachartikel in der Zeitschrift Briefings in Bioinformatics, online erschienen am 23.03.2026. Verfasst von Yiran Song und 6 weiteren Autor:innen. Beteiligte Einrichtung: Division of Computational Health Sciences, University of Minnesota, Mayo D528, 420 Delaware St SE, Minneapolis, MN 55455, United States. Hauptthemen laut MEDLINE: Neoplasms. Die Arbeit wurde bisher 5-mal zitiert (Crossref) und nennt 42 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: NIH HHS, NCCIH NIH HHS, NIA NIH HHS.
Abstract (Originalsprache)
Despite advances in multimodal cancer prognosis, robust performance in practical settings remains hindered by three critical barriers: ubiquitous data incompleteness, failure to model cancer-specific biology, and cross-institutional instability. We address these practical challenges through a cancer-type-aware framework that uniquely combines adaptive gated fusion for missing modalities, hybrid architecture for cancer heterogeneity, and demonstrated cross-institutional robustness. By establishing histopathology as the universally available anchor modality while adaptively incorporating RNA expression and clinical text through gated fusion, our framework maintains robust performance under realistic data constraints. Evaluation across 10 The Cancer Genome Atlas cancer types demonstrated superior performance (C-indices 0.578–0.778; mean 0.670 $\pm $ 0.066), with state-of-the-art results in six cancer types. The framework maintained predictive performance under missing data scenarios, with C-indices ranging from 0.621 to 0.627 for missing RNA and from 0.568 to 0.606 for missing clinical text. Cross-institutional validation across 12–38 centers demonstrated robust cross-institutional performance (standard deviations <0.040 in eight of 10 cancer types). This methodological framework addresses key technical prerequisites—handling missing data, modeling cancer heterogeneity, and ensuring cross-institutional stability—for multimodal survival prediction, providing computational foundations necessary for future prospective clinical validation.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Division of Computational Health Sciences, University of Minnesota, Mayo D528, 420 Delaware St SE, Minneapolis, MN 55455, United States.
- Department of Electrical and Computer Engineering, University of Minnesota, Minneapolis, MN 55455, United States.
- Advanced Imaging Research Center, Department of Biomedical Engineering, Peter O'Donnell Jr. Brain Institute, 2201 Inwood Road, NE3.218, Dallas, TX 75390, United States.
- Department of Population Health Sciences, Weill Cornell Medicine, New York, NY 10065, United States.
ORCID-Profile: Yiran Song, Yifan Peng, Mingquan Lin
Förderung
- NIH HHS
- NCCIH NIH HHS
- NIA NIH HHS
- NCI NIH HHS
- National Institutes of Health
- National Center for Complementary and Integrative Health
- National Institute on Aging
- National Cancer Institute
Zitieren
APA
Song, Y., Zhan, Z., Xie, F., Wang, N., Peng, Y., Zhang, R., & Lin, M. (2026). A cancer-type-aware framework for robust multimodal survival prediction under missing modalities. Briefings in Bioinformatics, 27(2). https://doi.org/10.1093/bib/bbag124
Vancouver
Song Y, Zhan Z, Xie F, Wang N, Peng Y, Zhang R, et al. A cancer-type-aware framework for robust multimodal survival prediction under missing modalities. Briefings in Bioinformatics. 2026;27(2). doi:10.1093/bib/bbag124
BibTeX
@article{Song2026,
author = {Song, Yiran and Zhan, Zaifu and Xie, Feng and Wang, Nian and Peng, Yifan and Zhang, Rui and Lin, Mingquan},
title = {A cancer-type-aware framework for robust multimodal survival prediction under missing modalities},
journal = {Briefings in Bioinformatics},
year = {2026},
volume = {27},
number = {2},
doi = {10.1093/bib/bbag124}
}
EUVIMED