🤖 AI Summary
Head and neck cancer involves high-dimensional, heterogeneous multimodal clinical data with complex temporal dynamics, posing significant challenges for existing survival prediction methods to effectively model structured clinical workflows. This work proposes a clinical pathway-guided heterogeneous hierarchical directed graph framework that represents a patient’s care trajectory as a temporally aware sequence aligned with key diagnostic steps. By leveraging a hierarchical topology and a heterogeneous message-passing mechanism, the model flexibly integrates multimodal information while robustly handling missing data and asymmetric cross-modal relationships. Evaluated on two public datasets, the proposed approach achieves state-of-the-art discriminative performance and well-calibrated predictions. Ablation studies further confirm the contribution of each component to the overall efficacy of the framework.
📝 Abstract
Accurate survival prediction is essential for personalized treatment planning in head and neck cancer, yet remains challenging due to the heterogeneous and high-dimensional nature of multimodal clinical data. While deep survival models have improved predictive performance over classical statistical approaches, existing methods typically rely on static fusion strategies or temporally agnostic modeling, limiting their ability to capture structured clinical workflows. In this work, we propose ChronoSurv, a heterogeneous hierarchical directed graph framework for multimodal survival analysis. ChronoSurv represents patient care as a progression-aware clinical trajectory using directed graphs aligned with key diagnostic steps. A hierarchical topology incorporates fine-grained, coarse, and global representations, further supporting flexible adaptation to missing modalities, while heterogeneous message passing models complex and asymmetric relationships across modalities and clinical steps. Experimental results on two public datasets demonstrate that ChronoSurv achieves state-of-the-art discriminative performance while maintaining statistically reliable calibration. Comprehensive ablation studies further confirm the contribution of each architectural component, highlighting the potential of trajectory-aware graph modeling for multimodal survival prediction.