DynSHAP: Towards Explainable Dynamic Survival Analysis
This research is important for anyone working on or evaluating AI models in healthcare, as it provides a method to make complex survival analysis models more transparent and trustworthy.
This paper introduces DynSHAP, a SHAP framework for dynamic survival analysis.
It addresses the black box nature of deep learning models in this field by extending SHAP estimators to handle longitudinal patient data and functional survival outputs.
The proposed Temporal DynSHAP method learns linear dependencies in features over time and uses conditional sampling for more accurate explanations.
When tested on real-world clinical datasets, DynSHAP provides attributions that are faithful to the model's learning, helping medical experts understand which patient information influenced predictions and when.
Deep learning models for dynamic survival analysis (DSA) achieve strong predictive performance by incorporating longitudinal patient data, but their black box nature limits clinical trust and adoption. Existing explainability methods cannot handle longitudinal, irregular inputs and functional survival outputs simultaneously, which limits their usability in DSA. We propose DynSHAP, a SHAP framework suited specifically for dynamic survival analysis. It extends common marginal SHAP estimators to this setting by treating time--feature pairs as players in the Shapley game. We further introduce Temporal DynSHAP, which learns linear dependencies in features over time and uses conditional sampling to address them in explanations. When applied to synthetic data with known ground-truth attributions, Temporal DynSHAP recovers temporally dependent features more accurately than marginal estimators for a given state-of-the-art model. Applied to two real-world clinical datasets and two DSA architectures, DynSHAP produces attributions faithful to model learning, allowing medical experts to see which patient information drove the prediction and when.
This report is categorized under AITOP's "Papers" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include DynSHAP, interpretability.
Source attribution: curated from arXiv cs.AI (@Nastasya Anokhina, Jonas Jürß, Pietro Liò).