FLUTE

The FLUTE project aims at researching a secure and privacy-preserving platform for federated machine learning, and apply this system to the use case of learning a predictive model to predict from prostate MRI images whether or not a biopsy is needed to detect clinically significant prostate cancer.  Hence, the goal is to train such predictive models while the data stays within the secured premises of the data owners and can at no point be revealed.

One of the partners of FLUTE i HL7-Europe, who supports the project for the use of FHIR and DICOM and medical data standards in general.

More information about FLUTE can also be found at https://fluteproject.eu/

Reason for applying to HSbooster.eu services

The project is interested in making specific contact with organisations working on AI standards: the project could contribute to proposing standard ways to represent federated learning pipelines and trustworthiness guarantees such as guarantees of privacy, security, fairness, performance etc. They are also interested in the relation to the regulatory compliance of systems, e.g., to understand whether and how such standards can help to determine whether systems satisfying some standardised criteria can more easily be identified as satisfying the high-level criteria regulation such as the GDPR or AI Act requires.  So in general the project is interested in contacts, understanding what already exists, understanding of how to contribute to standards.

Next to their interest in AI standardisation in general, the project is interested in AI in medicine in particular.  E.g., HL7 is interested in standardisation of data formats also include AI-related objects such as models or privacy-related information. In that direction, the project has a partner involved in the medical data standardisation process (e.g., FHIR), but would welcome making more contacts with partners interested in the standardisation of this domain.