🤖 AI Summary
To address the challenge of accurately estimating conditional average treatment effects (CATE) under distributed sensitive data—where centralized access to raw data is infeasible—this paper proposes the first privacy-preserving collaborative double machine learning (DML) framework. Our method integrates differential privacy with semiparametric modeling to enable non-iterative, communication-efficient CATE estimation and statistical inference without sharing raw data or intermediate gradients. A key innovation is the construction of an accumulative, cross-temporal and cross-institutional knowledge base that supports continual model evolution. Experiments on synthetic, semi-synthetic, and real-world healthcare datasets demonstrate that our approach matches or exceeds the estimation accuracy of existing methods, significantly improves robustness to model misspecification, and achieves superior privacy–utility trade-offs under constrained privacy budgets.
📝 Abstract
Estimation of conditional average treatment effects (CATEs) is an important topic in sciences. CATEs can be estimated with high accuracy if distributed data across multiple parties can be centralized. However, it is difficult to aggregate such data owing to confidential or privacy concerns. To address this issue, we proposed data collaboration double machine learning, a method that can estimate CATE models from privacy-preserving fusion data constructed from distributed data, and evaluated our method through simulations. Our contributions are summarized in the following three points. First, our method enables estimation and testing of semi-parametric CATE models without iterative communication on distributed data. Our semi-parametric CATE method enable estimation and testing that is more robust to model mis-specification than parametric methods. Second, our method enables collaborative estimation between multiple time points and different parties through the accumulation of a knowledge base. Third, our method performed equally or better than other methods in simulations using synthetic, semi-synthetic and real-world datasets.