Back to events

Past event

BrainTech Alliance in Paris

September 15, 2026 — Paris Brain Institute (ICM), Hôpital Pitié-Salpêtrière

Following the BrainTech Alliance's previous gatherings in Palo Alto, London and San Carlos, our 4th edition took place in Paris on September 15, 2026, at the Paris Brain Institute. Researchers, clinicians, industry leaders and investors met around one question: how are data and AI transforming neuro-innovation? After an opening by Pr Stéphanie Debette (Paris Brain Institute) and François Véron (BrainTech Alliance, Newfund Capital), Ira Hebold Haraldsen, AI-Mind Project Coordinator and BrainSymph Chair, set the tone with a keynote on how AI and data can turn brain health into a science of prediction and prevention. Three roundtables then followed the path from data to patients. The first lesson: more data does not mean better predictions. Neuroscience has never had so much of it, yet it still cannot tell which patients with mild cognitive impairment, about 40% of whom progress to dementia, will actually decline. For the speakers, the bottleneck is data quality rather than model power, since AI amplifies both the strengths and the weaknesses of what it is fed. Deep, longitudinal, standardized follow-up of the same individuals remains irreplaceable. "Size isn't enough," as Pr Jean-Christophe Corvol put it. And while the infrastructure exists, data still rarely moves between hospitals, platforms and countries, held back less by regulation than by institutional will and simple processes. The second lesson: think product from day one. AI-Mind, a Horizon 2020 project spanning eight European countries, was built to reach patients, not only to produce publications, and Ira Haraldsen warned against "well-funded cohorts built in isolation that never reach patients." The same logic runs through industry. Raidium is betting on foundation models and a single AI-native radiology viewer, because radiologists will not buy a thousand separately cleared tools. Neotiv turned a memory paradigm developed on 7T MRI into a smartphone test. And in brain tumor diagnosis, hundreds of hospitals already share data through Heidelberg's methylation classifier, proof that "governance came after practice." The third lesson concerned the market itself. Decades-old neurotech such as deep brain stimulation still reaches only a small share of eligible patients, and data alone will not fix adoption. Amy Kruse (Satori Neuro) warned founders against plans that need "four miracles in a row": biomarker discovery, implementation, clinician adoption and payment. The advice from the room was consistent: identify early who pays, prove the concept in small cohorts, and stay honest with partners. "Don't reshape your core problem to fit a meeting," said Thomas Trailov (Air Liquide), while Franck Augé (Servier) reminded founders that the first step toward pharma is simply reading its partnering page. One conviction ran through the whole afternoon: Europe's long-running cohorts and registries are a real competitive advantage, but a fragile one. Keeping it means making data flow, sharing costs between public and private actors, and helping companies scale without having to move to the US to raise their Series A. Many thanks to the Paris Brain Institute for hosting us, and to our speakers and moderators: Sylvie Bothorel, Thomas Nedelec, Géraldine Farjot, Pr Franck Bielle, Paul Hérent, Pr Emrah Düzel and Edward Ruppel, alongside those quoted above. The detailed insights from each session are available on the BrainTech Alliance Insights page.