MissHyper: Restoring Clinical Synchronicity in Missingness-Guided Hypergraph Forecasting

2026-07-23

AI Summary

Clinical multivariate time series often lose the synchronous contextual information of a patient’s state due to the isolated handling of observations at the same timestamp. To address this issue, this work proposes MissHyper, a novel model that, for the first time, restores snapshot-level synchronicity during the event initialization phase through a missingness-guided hypergraph structure. By integrating support density encoding with an adaptive gating mechanism, MissHyper effectively fuses local contextual information with node features. Extensive experiments on the PhysioNet 2012, MIMIC-III, and MIMIC-IV datasets demonstrate that the proposed method significantly outperforms existing hypergraph-based baselines, thereby validating the critical role of synchronous context modeling in sparse clinical time-series prediction.

Abstract

Clinical irregular multivariate time series are shaped not only by physiological dynamics but also by the measurement process that determines when and what to observe. In event-centric models, however, co-timestamp structure can be flattened too early: measurements acquired at the same timestamp are embedded as isolated nodes, leaving local patient-state context unavailable until later message-passing layers. We study this pre-propagation representation bottleneck and address it by restoring co-timestamp context before message passing begins. We propose MissHyper, a missingness-guided hypergraph forecasting model with pre-propagation synchronicity restoration. MissHyper augments each event with a local support-density cue, aggregates co-timestamp records to recover patient-state context, and uses a missingness-guided gate to adaptively fuse node-specific evidence with the recovered context. Across PhysioNet 2012, MIMIC-III, and MIMIC-IV, MissHyper achieves consistent gains in multi-step forecasting and outperforms a strong hypergraph baseline. These results suggest that improving event initialization can benefit sparse clinical forecasting without requiring a redesigned downstream propagation architecture. Ablations indicate that snapshot restoration, adaptive fusion, and support-density encoding all contribute, pointing to event initialization as a critical design axis for sparse clinical forecasting.

Problem

• clinical time series
• missingness
• hypergraph
• event-centric modeling
• synchronicity

Innovation

• missingness-guided
• hypergraph forecasting
• clinical time series
• synchronicity restoration
• event initialization