Brain Health Trajectories

The Brain Health Trajectories (BHT) project aims to revolutionize the screening of brain health by defining it as a state where the organ operates within specific, quantifiable parameter ranges. The project focuses on early identification of at-risk individuals through a battery of structural, functional, and model-based markers.

In 2025,  several scientific and technological milestones were achieved across the three Work Packages (WPs).

• WP1: Structural (Lesional) Biomarkers A major milestone was reached with the development of a toolkit for unsupervised anomaly detection using state-of-the-art generative diffusion neural networks. Originally optimized and tested on public datasets of stroke patients, these methods are now applied to the F-TRACT cohort, which contains over 1,300 intracranial electrophysiological and imaging datasets, serving as the primary engine for validating these tools and linking anatomical anomalies to physiological changes.

• WP2: Dynamical (Non-Lesional) Biomarkers The team has focused on moving beyond traditional functional connectivity by emphasizing dynamic and directed connectivity metrics to characterize how communication between brain regions changes over the human lifespan. A predictive model benchmark has been established to track patient-specific deviations from healthy trajectories.

• WP3: Model-Driven Biomarkers Significant advancements were made in "Virtual Twin" modeling with the first simulations incorporating dopaminergic neuromodulation. Furthermore, the Virtual Brain Inference pipeline was released, an open-source toolkit designed for efficient probabilistic inference on whole-brain models, bridging the gap between the mathematical models and clinical data.

The consortium produced 17 publications, including in the journal Brain and presentations at major international conferences such as NeurIPS 2025 and OHBM 2025. By integrating these results into the EBRAINS and harmonizing six datasets, BHT continues to strengthen France's strategic advantage in preventive and personalized brain medicine. The creation of the startup Karavela highlights the project’s translational success, utilizing brain foundational models to decode behavioral and cognitive information from brain activity for health monitoring.