VFCI: A Federated Framework for Vertical Causal Inference

Causal inference is crucial for understanding the relationship between cause and effect and is widely applied in fields such as medicine and policy decision-making. When the input data is distributed over multiple participants, most causal inference models require data to be centrally aggregated. However, participants are often prohibited or reluctant to share sensitive data due to data protection regulations and privacy concerns. To address this, federated learning has been proposed to perform data analysis in a privacy-preserving manner. Specifically, vertical federated learning is a subcategory where participants possess the same set of instances but different feature spaces. When causal inference meets vertical federated learning, a new topic arises: how to perform causal inference from multi-source vertically distributed data while protecting data privacy? In this paper, we propose a general framework, termed VFCI , for vertical federated causal inference. We design a general workflow that leverages a three-phase split mechanism to train traditional causal inference models securely. Based on this general workflow, we then show how to instantiate more complex causal inference models, such as two-stage approaches. Further, we conduct a comprehensive analysis regarding the security of our framework. Extensive experimental results verify the effectiveness and efficiency of VFCI over different workloads.