Last February, the French Data Protection Authority (CNIL) launched a call for innovative projects in the field of health data processing. With this initiative, the CNIL wanted to support innovative solutions in the field of health and give them the benefit of enhanced support in the context of a sandbox.The selected projects all have privacy issues at stake.

A project in a “sandbox” approach

The call for projects targets solutions that have not had the opportunity to be launched. The objective is to ensure that they can implement “privacy by design” at an early stage of development. The main selection criterion was to demonstrate an interest in privacy and data protection.

According to the Commission, the main objectives of the support provided to the projects are as follows:

“This support is based on the identification of difficulties to be overcome in a partnership manner during the experimentation. It will take the form of “close and repeated contacts with the CNIL’s legal and technical teams”: applicable rules, practical advice and an audit of the solutions developed by the project leader, in return for the commitments made by the latter”

The CNIL’s control services will have no knowledge of the projects chosen during the experimentation period, nor access to the files afterwards. The commission also specifies that it is not there to lift certain regulatory constraints, particularly with regard to the RGPD. The support will not aim to provide the project with an IT architecture or a test technique.

Between reinforced support and punctual support

Among the 60 applications submitted, the evaluation committee, made up of external experts and members of the CNIL, interviewed seven projects based on four criteria: the first being the interest for data protection, as mentioned above, and the other three involving the innovative nature of the project, the commitment of the project leader to the approach and the benefit for the public.

Four of them were selected to benefit from enhanced support:

• The project of the Lille University Hospital and the Magnet team at Inria concerning federated learning in artificial intelligence applied to clinical studies.
• The Resilience start-up project developing a solution to aid diagnosis in oncology.
• The Magellan project of the Clinityx research firm, which aims to build anonymous statistical indicators to describe populations in medical research;
• The Vertexa project of the Arras hospital centre, proposing a virtual reality solution for therapeutic purposes to combat eating disorders in minors.

Eight other projects will benefit from one-off support from the CNIL in the form of a personalised meeting:

• Mithril: use of secure microprocessor enclaves for confidentiality purposes.
• N4HCould from CEA: Cloud solution for health data with services.
• Takadoc: AI for the diagnosis and management of overweight.
• Sarus from Air Liquide: anonymization solution for the treatment of chronic diseases.
• MIR from AP-HP: imaging AI for decision support in oncology.
• A project led by the University of Rennes 1: development of a self-learning system with health data.
• SharpTx: AI for Alzheimer’s disease detection and preventive treatment application.
• Arkhn: a federated hospital data analysis tool.

In the coming months, all applicants will have access to webinars led by CNIL teams around health data.

Translated from La CNIL annonce les lauréats de son appel à projets autour de la protection des données de santé