Federated Causal Inference from Multi-Site Observational Data via Propensity Score Aggregation
arXiv:2505.17961v4 Announce Type: replace-cross Abstract: Causal inference typically assumes centralized access to individual-level data. Yet, in practice, data are often decentralized across multiple sites, making centralization infeasible due to privacy, logistical, or legal constraints. We address this problem by estimating the Average Treatment Effect (ATE) from decentralized observational data via a Federated Learning (FL) approach, allowing inference through the exchange of aggregate statistics rather than individual-level data. We propose a novel method to estimate propensity scores via
The increasing prevalence of multi-site data and growing privacy regulations necessitate new methods for data analysis without centralization, making federated causal inference a timely development.
This development addresses a critical bottleneck in leveraging disparate datasets for robust causal insights, especially in sensitive domains like healthcare or finance, broadening the application of AI and machine learning.
Traditional causal inference methods requiring centralized data are now augmented by federated approaches, enabling distributed causal analysis while preserving data privacy and logistical feasibility across various organizations.
- · Healthcare organizations
- · Financial institutions
- · Privacy-focused tech companies
- · Distributed research consortia
- · Companies reliant on centralized data collection
- · Traditional data brokers
- · Single-site research models
Propensity score aggregation allows for estimating Average Treatment Effects (ATE) without sharing raw individual data.
This framework could lead to a new standard for collaborative research and model training across privacy-sensitive domains.
The broader adoption of federated causal inference may accelerate the development of agentic systems that learn from diverse, protected data sources without ever directly accessing them.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI