Premium
Modelling and estimation for optimal treatment decision with interference
Author(s) -
Su Lin,
Lu Wenbin,
Song Rui
Publication year - 2019
Publication title -
stat
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.61
H-Index - 18
ISSN - 2049-1573
DOI - 10.1002/sta4.219
Subject(s) - estimator , interference (communication) , computer science , outcome (game theory) , regimen , artificial intelligence , machine learning , mathematical optimization , mathematics , computer network , statistics , medicine , channel (broadcasting) , mathematical economics
In many network‐based intervention studies, treatment applied on an individual or his or her own characteristics may also affect the outcome of other connected people. We call this interference along network. Approaches for deriving the optimal individualized treatment regimen remain unknown after introducing the effect of interference. In this paper, we propose a novel network‐based regression model that is able to account for interaction between outcomes and treatments in a network. Both Q‐learning and A‐learning methods are derived. We show that the optimal treatment regimen under our model is independent from interference, which makes its application in practice more feasible and appealing. The asymptotic properties of the proposed estimators are established. The performance of the proposed model and methods is illustrated by extensive simulation studies and an application to a mobile game network data.