Statistical analysis of patient data can help evaluate health risks, determine best practices, and develop new treatments that can alter the quality of life for millions of people. However, privacy concerns make it infeasible to provide researchers with unfettered access to medical records. The practical previous approaches to this problem have tried to de-identify the data by removing personally identifiable information from the databases, but recent research demonstrates that this provides insufficient protection. The concept of differential privacy promises a practical resolution to this conflict. This project is creating practical differentially-private algorithms for common data analysis and publication methods, and validating these with the data analysis tasks of a new medical study. For more information visit the Enabling Biomedical Research with Differential Privacy web page.
Enabling Biomedical Research with Differential Privacy
permalink. Both comments and trackbacks are currently closed.. Bookmark the