We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

stat.ML

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Statistics > Machine Learning

Title: A Tree-based Federated Learning Approach for Personalized Treatment Effect Estimation from Heterogeneous Data Sources

Abstract: Federated learning is an appealing framework for analyzing sensitive data from distributed health data networks. Under this framework, data partners at local sites collaboratively build an analytical model under the orchestration of a coordinating site, while keeping the data decentralized. While integrating information from multiple sources may boost statistical efficiency, existing federated learning methods mainly assume data across sites are homogeneous samples of the global population, failing to properly account for the extra variability across sites in estimation and inference. Drawing on a multi-hospital electronic health records network, we develop an efficient and interpretable tree-based ensemble of personalized treatment effect estimators to join results across hospital sites, while actively modeling for the heterogeneity in data sources through site partitioning. The efficiency of this approach is demonstrated by a study of causal effects of oxygen saturation on hospital mortality and backed up by comprehensive numerical results.
Comments: An earlier version won JSM 2021 Student Paper Competition (SLDS section) Honorable Mention
Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME)
Cite as: arXiv:2103.06261 [stat.ML]
  (or arXiv:2103.06261v2 [stat.ML] for this version)

Submission history

From: Xiaoqing Tan [view email]
[v1] Wed, 10 Mar 2021 18:51:30 GMT (1002kb,D)
[v2] Sun, 18 Apr 2021 00:35:12 GMT (2264kb,D)

Link back to: arXiv, form interface, contact.