Dynamic Treatment on Networks

2026-05-07

AI Summary

This work addresses the limitations of existing social network intervention methods, which often neglect network structure or rely solely on static strategies, thereby failing to effectively harness spillover effects. The authors propose Q-Ising, a three-stage framework that first estimates adoption dynamics from single-panel data using a Bayesian dynamic Ising model, then enriches treatment history through continuous posterior latent states, and finally optimizes dynamic intervention policies via offline reinforcement learning. This approach uniquely integrates dynamic intervention under network interference with explicit structural awareness, yielding interpretable policies that quantify uncertainty and providing a finite-sample regret upper bound accounting for both network abstraction and state estimation error. Experiments on microfinance diffusion in Indian villages and heterogeneous SIS stochastic block models demonstrate that its adaptive targeting significantly outperforms static centrality-based benchmarks.

Abstract

In networks, effective dynamic treatment allocation requires deciding both whom to treat and also when, so as to amplify policy impact through spillovers. An early intervention at a well-connected node can trigger cascades that change which nodes are worth targeting in the next period. Existing treatment strategies under network interference are largely static while dynamic treatment frameworks typically ignore network structure altogether. We integrate these perspectives and propose Q-Ising, a three-stage pipeline that (i) estimates network adoption dynamics via a Bayesian dynamic Ising model from a single observed panel, (ii) augments treatment adoption histories with continuous posterior latent states, and (iii) learns a dynamic policy via offline reinforcement learning. The Bayesian mechanism enables uncertainty quantification over dynamic decisions, yielding posterior ensemble policies with interpretable spillover estimates. We provide a finite-sample regret upper bound that decomposes into standard offline-RL uncertainty, network abstraction error, and first stage error in Ising state estimation. We apply our method to data from Indian village microfinance networks and synthetic stochastic block models under simulated heterogeneous susceptible-infected-susceptible (SIS) dynamics and demonstrate that adaptive targeting outperforms static centrality benchmarks.

Problem

• dynamic treatment
• network interference
• spillover effects
• treatment allocation
• policy impact

Innovation

• dynamic treatment allocation
• network interference
• Bayesian Ising model
• offline reinforcement learning
• spillover effects