I think it would be a good idea for organizations that collect sensitive data but only need to leverage its aggregate properties to at least try to protect individuals through random perturbation. It may still be an imperfect solution, but it would at least be a good-faith step in the right direction.

For background: Over 87% of the population can be identified if you have just 3 pieces of information. Their zip code, date of birth and gender but there are ways to use privacy protecting algorithms for large population sets for research while still protecting individual’s information.

Latanya Sweeney’s classic study she was able to identify Governor William Weld by matching masked medical data from the Group Insurance Commission of Massachusetts with publicly available voters lists is a foreshadowing of what can happen.

There has been some really interesting research done “Privacy Preserving Data Analysis” which is based on the work of Microsoft Research (MSR) researchers Cynthia Dwork and Frank McSherry (as well as S Chawla, K Talwar, A Blum, K Nissim, and A Smith) .

The basic premise of their research is that the addition of noise (is shift the zip code by one, the age by one) to the data will be able to protect the individuals underneath the aggregate data and still produce large scale research results.

For a deatialed overview of the actual algorithm check out Denny Lee’s blog at Microsoft .

http://denster.spaces.live.com/

Sherry Reynolds
Alliance4Health