Federated Synthetic Biomedical Data Generation
A research concept for enabling institutions to develop synthetic biomedical datasets collaboratively without centralising their raw patient data.
Explore collaborative research without bringing all raw patient records into one place.
Institutions may share a research question while holding data under different responsibilities and access restrictions. This concept explores collaborative creation of synthetic biomedical datasets while retaining institution-level control of raw records.
- Institution-level data control
- conditional synthetic profiles
- collaborative learning
- audit-oriented records
- adaptable deployment across local and hosted environments
- export for authorised research workflows
The user experience
Participating teams define an authorised research purpose, coordinate the study and review generated profiles. Audit-oriented records and controlled exports are intended to support accountable use of the resulting material.
Practical value
Supports research collaboration as a design goal. Synthetic data is not automatically anonymous, representative or clinically valid and requires independent review.
Application scenarios
Examples of potential use, not claims of an existing deployment.
Multi-institution research
Explore data-generation collaboration between universities and healthcare research groups that cannot simply pool their raw records.
Study preparation
Investigate synthetic profiles for early experimentation before seeking approval for a specific downstream study.