Gradient-based Explanations for Deep Learning Survival Models

2025-02-07

AI Summary

Deep learning survival models achieve high predictive accuracy but suffer from limited clinical trustworthiness due to their “black-box” nature. To address time-to-event prediction, this paper introduces the first gradient-based interpretability framework for survival analysis, centered on Time-Aware GradSHAP(t)—the first method integrating gradient backpropagation with SHAP theory in survival modeling, supporting multimodal inputs and dynamic temporal attribution. Theoretically, we establish differentiability and consistency guarantees for survival gradient explanations; practically, we propose a standardized visualization paradigm. Extensive evaluation on synthetic and real-world clinical datasets demonstrates that GradSHAP(t) significantly outperforms SurvSHAP(t) and SurvLIME, achieving a 4.7% AUC improvement and 3.2× speedup in computation, while preserving both local and global interpretability and accurately capturing time-varying feature importance.

Technology Category

• Machine Learning: Transparent, Interpretable, Explainable ML
• Humans and AI: Explainable AI (XAI) for Human Understanding
• Computer Vision: Interpretability, Explainability, and Transparency

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Security and Privacy: Data transparency and provenance

Abstract

Deep learning survival models often outperform classical methods in time-to-event predictions, particularly in personalized medicine, but their"black box"nature hinders broader adoption. We propose a framework for gradient-based explanation methods tailored to survival neural networks, extending their use beyond regression and classification. We analyze the implications of their theoretical assumptions for time-dependent explanations in the survival setting and propose effective visualizations incorporating the temporal dimension. Experiments on synthetic data show that gradient-based methods capture the magnitude and direction of local and global feature effects, including time dependencies. We introduce GradSHAP(t), a gradient-based counterpart to SurvSHAP(t), which outperforms SurvSHAP(t) and SurvLIME in a computational speed vs. accuracy trade-off. Finally, we apply these methods to medical data with multi-modal inputs, revealing relevant tabular features and visual patterns, as well as their temporal dynamics.

Problem

• Explain deep learning survival models
• Extend gradient-based explanation methods
• Visualize temporal feature effects

Innovation

• Gradient-based explanation for survival models
• Extends SHAP and LIME in survival analysis
• Visualizes temporal dynamics in feature effects