Elisa Ficarra

Abstract Accurate survival prediction in non-small cell lung cancer (NSCLC) requires integrating clinical, radiological, and histopathological data. Multimodal deep learning (MDL) can improve precision prognosis, but small cohorts and missing modalities limit its clinical applicability, as conventional approaches enforce complete-case filtering or imputation.

The results demonstrate that the presented graph-based framework consistently outperforms standard MIL baselines across all evaluated tasks. The gains are particularly notable in HR status prediction, where attention-based integration of multiple tissues appears to be critical. This likely stems from the inability of single-site models to capture morphologic patterns that are distributed across different anatomical contexts.

Keywords pediatric cancer medulloblastoma multiomics integration proteomics phosphoproteomics transcriptomics metabolomics lipid biosynthesis lipid droplets lipid oxidative stress Get full text access Log in, subscribe or purchase for full access. References 1. Louis, D.N. ∙ Perry, A. ∙ Reifenberger, G. ... The 2016 World Health Organization Classification of Tumors of the Central Nervous System: A Summary Acta Neuropathol. 2016; 131:803-820 2. Ramaswamy, V. ∙ Remke, M. ∙ Bouffet, E. ...