Predicting counterfactual distributions in complex dynamical systems is essential for scientific modeling and decision-making in domains such as public health and medicine. However, existing methods often rely on point estimates or purely data-driven models, which tend to falter under data scarcity. We propose a time series diffusion-based framework that incorporates guidance from imperfect expert models by extracting high-level signals to serve as structured priors for generative modeling. Our method, ODE-Diff, bridges mechanistic and data-driven approaches, enabling more reliable and interpretable causal inference. We evaluate ODE-Diff across semi-synthetic COVID-19 simulations, synthetic pharmacological dynamics, and real-world case studies, demonstrating that it consistently outperforms strong baselines in both point prediction and distributional accuracy.
@article{arxiv.2508.13355,
title = {Counterfactual Probabilistic Diffusion with Expert Models},
author = {Wenhao Mu and Zhi Cao and Mehmed Uludag and Alexander Rodríguez},
journal= {arXiv preprint arXiv:2508.13355},
year = {2025}
}