Decentralized differentially-private methods for dynamic data release and analysis

项目总结

PROJECT SUMMARY Data sharing and information exchange are playing critical roles in biomedical data science to improve quality of care, accelerate discovery, and promote meaningful secondary use of clinical data. But privacy is a big concern to the public. Existing distributed data analysis methods do not address the security and privacy issues in exchanging intermediary statistics and they cannot handle dynamic database updates very well. This project aims at designing and implementing differentially-private decentralized methods for dynamic data dissemination and analysis. We plan to use genomic and clinical data from both public domain and local institutions (UCSD and Emory) to carefully evaluate the feasibility and efficiency of our proposed new methods.