SL-BiLEM: Structured Learnable Behavior-in-the-Loop Epidemic Modeling for Forecasting and Policy Evaluation

2026-05-26

AI Summary

This study addresses the challenge of distribution shift in epidemic forecasting caused by feedback loops between human behavior and disease transmission, which undermines data-driven models. The authors propose a structured, learnable transmission model that decomposes the effective transmission rate into a product of a baseline infectivity rate and modulatory factors capturing the effects of policies, media, and behavioral compliance. For the first time in behavioral response modeling, they incorporate physically grounded constraints—monotonicity, smoothness, and bounded jumps—by embedding domain knowledge into a regularized neural network. The approach substantially improves generalization: on real-world data, it achieves a 76% gain in predictive performance over neuro-mechanistic baselines and exhibits only 53% out-of-distribution degradation (versus 1142% for the baseline). In 27 synthetic counterfactual experiments, it attains 100% confidence interval coverage and treatment effect accuracy exceeding 0.85.

Abstract

Epidemic forecasting faces a fundamental challenge: human behavior dynamically responds to disease spread, creating feedback loops that induce distribution shifts at policy intervention points. This renders data-driven models unreliable under distribution shift. We propose \textbf{SL-BiLEM} (Structured Learnable Behavior-in-the-Loop Epidemic Model), leveraging physical constraints as regularization for robust extrapolation. The framework decomposes effective transmission as $β_{\text{eff}}(t,g) = β_0(g) \times m_{\text{policy}}(t) \times m_{\text{media}}(t) \times m_{\text{comp}}(t,g)$, where monotonicity, smoothness, and bounded-jump constraints on the learned compliance function maintain predictive validity under novel policy regimes. Beyond forecasting, SL-BiLEM enables counterfactual analysis for intervention decision support. We validate forecasting on three real-world datasets (cruise ship, school influenza, and school-district COVID-19 surveillance) and evaluate counterfactual recovery on synthetic benchmarks with known ground truth. SL-BiLEM demonstrates: (1) 76\% improvement over neural-mechanistic baselines, with only 53\% OOD degradation versus 1142\% for neural baselines under policy-induced shift; (2) 100\% bootstrap CI coverage across 27 synthetic counterfactual experiments; and (3) Treatment Effect Accuracy exceeding 0.85. These results establish SL-BiLEM as an interpretable tool for public health decision-makers seeking accurate prediction and principled intervention planning.

Problem

• epidemic forecasting
• human behavior
• distribution shift
• policy intervention
• feedback loops

Innovation

• behavior-in-the-loop
• distribution shift robustness
• structured regularization
• counterfactual policy evaluation
• epidemic forecasting