To allow a freer use of personal data in medicine, the startup Octopize develops its Avatar method. A reliable technology capable of completely anonymizing data.
What is Avatar?
Published on March 10 in an article in Nature Digital Medicine, Octopize, in collaboration with the University Hospital of Nantes, Nantes University, INSERM and École Centrale Nantes, explains how Avatar works.
On the market, many methods claim to be able to anonymize personal data. In reality, these data are only pseudonymized and run the risk of being re-identified. The Avatar method completely anonymizes the data and therefore avoids the risk of re-identification. By drawing on the individual’s data, the Avatar algorithm generates new synthetic data centered on the owner. The result is an anonymized avatar created by specific methods that do not exist in other anonymization methodologies.
How do we ensure that the data is anonymized?
To ensure privacy, Avatar meets three criteria of the European Data Protection Committee.
The first is individualization. Avatars datasets are individualizable, so you can’t isolate an individual. The second criterion is correlation; Avatar prevents two datasets about the same group of individuals from being linked. The last point is the inference. This process blocks the ability to infer sensitive information about an individual.
With Avatar’s non-personal data, owners are free to publish their datasets in open data, or to use them. This data can then be used to train AI models, contribute to researching and statistical studies in medicine and many other fields.