Data & Knowledge Systems · Data & Knowledge Systems Lab

Federated Learning for Privacy-Preserving Analytics

Hospitals and clinics often can't share patient data directly, even for research that could help their own patients. This project's federated learning framework trains shared models across institutions without the raw data ever leaving its source. Two partner hospital research networks are using an early version in production.

Research Question

Hospitals and clinics often can't share patient data directly, even for research that could help their own patients.

Project Timeline
Concept
2022-08
Investigation
2023-01
Prototype
2023-06
Evaluation
2023-11
Publication
2024-04
Deployment
2024-09
Student Involvement

4 graduate researchers; results co-authored with two external hospital research partners.

Lead Researcher
Priya Raghunathan
Research Group

Data & Knowledge Systems Lab

Started

August 1, 2022

Current Stage

Publication