Quantifying Cross-Modal Interactions in Multimodal Glioma Survival Prediction via InterSHAP: Evidence for Additive Signal Integration

TL;DR

This study uses InterSHAP to quantify cross-modal interactions in glioma survival models, revealing that higher predictive performance correlates with more additive, rather than synergistic, signal integration.

Contribution

It adapts InterSHAP for Cox models and demonstrates that improved survival prediction arises from additive signal combination, challenging the assumption of beneficial synergy.

Findings

Higher-performing models show lower cross-modal interaction.

Additive contributions dominate over interaction effects.

Performance improvements are due to complementary signals, not synergy.

Abstract

Multimodal deep learning for cancer prognosis is commonly assumed to benefit from synergistic cross-modal interactions, yet this assumption has not been directly tested in survival prediction settings. This work adapts InterSHAP, a Shapley interaction index-based metric, from classification to Cox proportional hazards models and applies it to quantify cross-modal interactions in glioma survival prediction. Using TCGA-GBM and TCGA-LGG data (n=575), we evaluate four fusion architectures combining whole-slide image (WSI) and RNA-seq features. Our central finding is an inverse relationship between predictive performance and measured interaction: architectures achieving superior discrimination (C-index 0.64$\to$0.82) exhibit equivalent or lower cross-modal interaction (4.8\%$\to$3.0\%). Variance decomposition reveals stable additive contributions across all architectures (WSI${\approx}$40\%, RNA${\approx}$55\%, Interaction${\approx}$4\%), indicating that performance gains arise from complementary signal aggregation rather than learned synergy. These findings provide a practical model auditing tool for comparing fusion strategies, reframe the role of architectural complexity in multimodal fusion, and have implications for privacy-preserving federated deployment.