Oral Presentations

The Feasibility of Garbled Circuits for Cross-Site Clinical Data Analytics

4:03 PM–4:21 PM Mar 24, 2020 (America - Chicago)



Programmatic Theme: Clinical Research Informatics

Abstract: We report preliminary results using a leading-edge garbled circuit framework to prototype secure computation protocols for clinical data analytics. Traditionally viewed as too difficult for non-experts and impractical to use, garbled circuits are fast becoming viable tools. We present a simple use case, which still requires an honest broker in current practice. Our results demonstrate that a framework like the EMP-toolkit are easy to onboard while offering strong security and reasonable performance.

Learning Objective: 1. Understand the limitations of current approaches used in cross-site clinical data analytics.
2. Understand the basics for garbled circuit, secure two party computation and oblivious transfers.
3. Understand why semi-honest is the threat model in this context.
3. Learn the basics for garbled circuit compiler, and how to use it to fast-prototype security protocols.


Xiao Dong (Presenter)
University of Illinois at Chicago

David Randolph, University of Illinois at Chicago
Xiao Wang, Northwestern University

Keywords, Themes & Types