Advancing Health Research While Safeguarding Genetic Privacy
MIE Assistant Professor Xubo Yue, in collaboration with Brown University, has received a $1M NSF grant for the project titled “SCH: Protecting Privacy and Promoting Fairness in Advanced Genomic Research using Federated Learning.” This project represents a significant step forward in advancing federated learning for large-scale, privacy-preserving analysis of biological data across multiple institutions – enabling collaboration without the need for data sharing.
Advances in sequencing technologies have enabled the generation of large-scale, multi-cohort genomic and genetic datasets. However, integrating data across institutions remains a major challenge due to privacy concerns, infrastructure disparities, and regulatory constraints. Centralized data sharing often poses risks, as genetic data can reveal sensitive information about individuals’ traits, health, and ancestry.
This project introduces FLAG (Federated Learning for Advanced Genomics), a federated learning framework that enables secure, collaborative analysis of genomic and genetic data without requiring data to be shared. The research includes three aims: (1) develop privacy-preserving representation learning methods to extract key molecular features across institutions; (2) design federated Bayesian models to improve the accuracy and generalizability of genetic risk predictions across diverse cohorts; and (3) release an open-source software platform to support decentralized analysis, making federated learning accessible even to institutions with limited resources.
FLAG addresses key barriers in multi-institutional collaboration by ensuring data confidentiality while supporting scalable and reproducible analysis. The framework is broadly applicable across biomedical domains, including genomics, electronic health records, and digital pathology. This work will equip the research community with practical tools for privacy-aware discovery and accelerate progress in precision medicine.