DBRNet: Advancing Individual-Level Continuous Treatment Estimation through Disentangled and Balanced Representation

Abstract

Estimating the individual-level continuous treatment effect holds significant practical importance in various decision-making domains, such as personalized healthcare and customized marketing. However, current methods for individual treatment effect estimation are limited to discrete treatments or rely on a simplistic approach of balancing the entire representation, which may lead to inaccurate estimation. To the best of our knowledge, no existing efforts is capable of precisely adjusting for selection bias in continuous settings. Hence, in this paper, we propose a novel Disentangled and Balanced Representation Network (DBRNet) for estimating the individualized dose-response function (IDRF), which learns disentangled representations and precisely adjusts for selection bias. Extensive results on synthetic and semi-synthetic datasets demonstrate that our DBRNet outperforms most state-of-the-art methods. Our code is avaiable at https://anonymous.4open.science/r/DBRNet_final_2-2B76.