Evolution With Purpose: Hierarchy-Informed Optimization of Whole-Brain Models

Researchers developed a new AI method called Hierarchy-Informed Optimization (HICO) that evolves more accurate whole-brain models. Unlike standard evolutionary approaches that overfit to individual MRI data, HICO uses biological knowledge about brain region hierarchy to guide the optimization of 20 key parameters. This approach was the only one that successfully generalized to new subjects and, crucially, used the model parameters to predict the subjects' real-world behavioral abilities.