Failure Modes of Time Series Interpretability Algorithms for Critical Care Applications and Potential Solutions
This addresses the challenge of deploying interpretable deep learning models in critical care, where patient survival depends on accurate and consistent predictions over time, though it appears incremental by building on existing interpretability methods.
The paper tackled the problem of interpretability algorithms failing in dynamic time series prediction tasks for critical care, such as issues with time-varying dependencies and temporal smoothness, and proposed learnable mask-based frameworks as solutions that incorporate constraints for more reliable interpretations.
Interpretability plays a vital role in aligning and deploying deep learning models in critical care, especially in constantly evolving conditions that influence patient survival. However, common interpretability algorithms face unique challenges when applied to dynamic prediction tasks, where patient trajectories evolve over time. Gradient, Occlusion, and Permutation-based methods often struggle with time-varying target dependency and temporal smoothness. This work systematically analyzes these failure modes and supports learnable mask-based interpretability frameworks as alternatives, which can incorporate temporal continuity and label consistency constraints to learn feature importance over time. Here, we propose that learnable mask-based approaches for dynamic timeseries prediction problems provide more reliable and consistent interpretations for applications in critical care and similar domains.