Time-series model explanations using meaningful regions and realistic baselines
GRAIL: From Saliency to Semantics in Time-Series Explanation
Artificial Intelligence
Summary
Interpreting how models make decisions from time-series data is challenging because important patterns span multiple time steps, not just one. The authors propose a method called GRAIL that explains model predictions by focusing on meaningful intervals, such as parts of an ECG heartbeat, rather than individual points in time. They generate realistic baselines for comparison using advanced techniques, which helps produce clearer and more trustworthy explanations. This approach can also include static patient information alongside time-based data for a fuller understanding.
What this means in practice
- •For clinical data scientists: Explain patient ECG-based model predictions using interpretable intervals alongside patient demographic data.
- •For time-series model engineers: Improve debugging and validation of time-series models by attributing predictions to domain-informed temporal regions rather than single points.
Authors
Enver Menadjiev, Yiqiu Shen, Jaesik Choi, Julia Stoyanovich
Abstract
Explaining a time-series model requires both locating the evidence and extracting meaningful knowledge from it. Yet a single timestep is not a semantic unit: the meaningful patterns live across many samples, not in any one of them. Per-timestep saliency is therefore noisy, high-dimensional, and unstable. Existing methods address this by introducing fixed-size windows and treating them as independent players in a cooperative game. Evaluating a coalition of such players requires defining a ``baseline'' region, which prior work tackles with fixed counterfactuals (e.g., a flat line, noise, or a spectral surrogate). This choice frequently dominates the resulting explanation, obscuring the model's true decisions. To address these issues, we propose GRAIL, a region-based time-series explanation framework that uses domain-specific intervals, such as the P-wave and the QRS complex of an electrocardiogram (ECG) beat, as players. We describe each region with human-readable features and generate counterfactual baselines with an on-manifold conditional generator. Such a choice of players also admits non-temporal features: static covariates enter as singleton players attributed with the temporal regions on one scale, so a patient's tabular evidence (e.g., age, sex) is weighed directly against the signal. We attribute the predictions to the features via Owen values with common-random-number variance reduction. Experimental results on both synthetic and real data show that the proposed framework produces faithful and robust explanations, evaluated under matched on- and off-manifold operators, offering meaningful knowledge to domain experts.