PASTEUR2030
NEURODEV project - Institut Pasteur and AP-HP, selected in the new "data challenges in health" call under the France 2030 program
As part of its France 2030 plan, the French government has launched a strategy to accelerate digital health, paving the way for France to become a leader in the field. Since 2020, particular emphasis has been placed on the use of artificial intelligence to process health data, for the benefit of patients.
One strand is to support the organization of "data challenges in health." These international challenges are data science competitions hosted on specialized platforms, aimed at addressing specific medical issues through the analysis of high-quality anonymized datasets made available to all participants. The participants are tasked with responding to the question by developing algorithms within a set time, based on the data provided.
The latest "data challenges in health" call for proposals, operated by Bpifrance in partnership with the Health Data Hub, was launched last year, and on July 10 ten new winning proposals were announced. The topics covered – ranging from cancer and cardiovascular diseases to neurodevelopmental disorders and critical care – all represent challenges for AI in health and reflect the same ambition, namely to speed up the development of AI solutions capable of improving diagnostics, prevention and treatment for patients. The projects will receive France 2030 funding and support from the Health Data Platform to organize their competition and pursue applications based on the results.
The ten selected projects include NEURODEV - Institut Pasteur and the Paris Public Hospital Network (AP-HP). At the Institut Pasteur, the project is coordinated by Roberto Toro, head of the Applied and Theoretical Neuroanatomy five-year group (NAAT), along with Katja Heuer and Nicolas Traut, specialists in neuroimaging, open science, and data science; in collaboration with Shauna Lee Katz of DARRI. The IT Department, led by Stéphane Fournier, will contribute to the deployment of the IT infrastructure. Teams at Robert-Debré Hospital, led by Prof. Richard Delorme, will provide their clinical and pediatric imaging expertise, particularly in the field of neurodevelopmental disorders.
The project aims to develop and evaluate artificial intelligence and machine learning algorithms capable of identifying, based on brain MRI scans, signatures associated with neurodevelopmental disorders. It will focus in particular on autism spectrum disorders and attention-deficit/hyperactivity disorder (ADHD), with the goal of eventually covering the full range of neurodevelopmental disorders. Ultimately, this research could help improve early detection, clinical assessment, and understanding of the underlying brain mechanisms involved.
NEURODEV builds on the experience of the previous IMPAC challenge, organized by the NAAT team at the Pasteur Institute in collaboration with Inria. That international challenge utilized brain imaging data from individuals with and without a diagnosis of autism spectrum disorder. Following a competition phase, the top teams participated in a collaborative hackathon at the Pasteur Institute. This approach made it possible to combine the most effective methods and led to a scientific publication in the journal NeuroImage.
The new challenge will draw on large, open-access international databases, as well as private datasets—including, notably, data from Robert-Debré Hospital—reserved for the independent evaluation of algorithms. The goal is to bring together more than 10,000 structural and functional MRI datasets, compared to approximately 2,000 in the previous challenge. The data available to participants will be anonymized and prepared in a format that facilitates its reuse by the scientific community.
In line with the principles of open science, the algorithms developed as part of the challenge will be shared as open source. The project will also enable the establishment of a data challenge platform at the Pasteur Institute, which can be used in the future for other research projects on campus.