Skip to content
Open access

Climate-Aware Self-Retrospective Representation Learning for Spatio-Temporal Epidemic Forecasting

Aug 2026 · Tropical Medicine and Infectious Disease · Vol 11, pp. 240 · 0 citations · 42 references
Medicine

TL;DR

This paper proposes CASRL, a Climate-Aware Self-Retrospective Representation Learning network for stable and meteorological-context-aware spatio-temporal epidemic forecasting, which produces forecasts with improved predictive stability at later forecast horizons.

Abstract

Spatio-temporal epidemic forecasting aims to predict future outbreak trajectories across interconnected regions from historical epidemiological observations and meteorological covariates. However, existing approaches often fail to preserve historically salient epidemic states or to fully exploit delayed and region-varying meteorological associations, leading to unstable temporal representations and insufficient meteorological-context-aware spatio-temporal context for prediction at later forecast horizons. In this paper, we propose CASRL, a Climate-Aware Self-Retrospective Representation Learning network for stable and meteorological-context-aware spatio-temporal epidemic forecasting. CASRL first employs a Self-Retrospective Epidemic Encoder (SREE) to retrospectively aggregate historically salient epidemic states through query-guided weighting and adaptive gating, thereby preserving informative historical epidemic states within the look-back window. It then introduces a Climate-Adaptive Graph Message Passing (CAGMP) module that breaks away from traditional passive feature concatenation. Instead, it constructs a separate meteorological-view predictive graph conditioned on the static spatial prior and adaptively fuses it with the incidence-associated topology to model complex cross-regional predictive associations. By integrating self-retrospective epidemic representations with meteorological-view spatio-temporal interactions, CASRL produces forecasts with improved predictive stability at later forecast horizons. Extensive experiments on two public influenza benchmarks show that CASRL is competitive at shorter forecast horizons and provides clearer advantages at later forecast horizons, particularly in phase-alignment-related evaluation and 15-week-ahead forecasting.

Read PDF

Similar papers

Book Open access Nov 2026

Graph Neural Delay Differential Equations for Continuous Epidemic Forecasting

Reliable epidemic forecasting is essential for early warning and proactive public health decision-making. Existing spatio-temporal forecasting methods often model regional interactions on fixed discrete grids and rely on instantaneous or uniformly delayed transmission assumptions, limiting their ability to capture asyn...

Ming-Da Xu, Z. Du, Chao Gao · 0 citations
Open access Nov 2026

Residual Diffusion-Enhanced Spatio-Temporal Network for Human Mobility Forecasting in Disaster Scenarios

Forecasting human mobility during disasters is essential for emergency response but remains challenging because disasters cause abrupt, uncertain, and context-dependent deviations from regular mobility patterns. Deterministic models capture dominant spatio-temporal patterns but tend to underrepresent these deviations,...

Xing Gao, Meng Chen, Nan Yin et al. · 0 citations
#machine learning Preprint Oct 2026

Learning and Forecasting Multiscale Spatiotemporal Epidemic Dynamics with Lagrangian-Eulerian Graph Neural ODEs

Accurate medium- to long-range epidemic forecasting requires extrapolating high-dimensional, heterogeneous and nonstationary dynamics from partial observations. Yet hybrid models often leave implicit which observations inform which epidemic processes and how learned graph interactions modify system state. We introduce...

Jin-Ying Duan, Qian Li, Biao Tang · 0 citations

More Sensors Only One Field: Rethinking Continual Spatio-Temporal Forecasting

Continual spatio-temporal forecasting supports traffic management and environmental monitoring under evolving dynamics and expanding sensor networks. However, conventional graph-based continual learning methods tie forecasting representations to the current sensor layout, so sensor expansion can alter the representatio...

Le-Wei Xie, Hao-Yu Zhang, Jia-Jun Zhou et al. · 0 citations
Preprint Oct 2026

Data-driven stochastic reconstruction of dengue epidemic dynamics: state-dependent stochasticity and temporal organization

Dengue epidemics exhibit strong stochastic variability, temporal memory and complex temporal organization that are difficult to capture using conventional deterministic epidemic models. However, the dependence of epidemic fluctuations on the underlying dynamical state and their influence on reconstruction across tempor...

Tongbram Surkumar Singh, Mairembam Kelvin Singh, Athokpam Langlen Chanu et al. · 0 citations
Open access Oct 2026

Real-Time Epidemic Forecasting Across Heterogeneous Populations: A Multi-Source Volatility-Gated Ensemble Framework

Accurate epidemic forecasting at a global scale is inherently challenging due to heterogeneous reporting quality, behavioural variability, and differing rates of policy changes across countries. This study presents a comprehensive forecasting pipeline trained on multi-source COVID-19 data spanning among 253 countries f...

Vithya Ganesan, V. A. Devi, Dr. Raju Shanmugam et al. · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.