REnaissance: Interventional Whole-Brain Modeling for Algorithmic Neuropsychiatry

REnaissance proposes an experimental resource for testing whether individualized brain models can explain differences in depression treatment response and ultimately improve care.

People with similar depression scores may differ in how they form expectations, evaluate outcomes, and choose actions. Algorithmic neuropsychiatry makes these processes explicit computational hypotheses and links them to brain activity, behavior, and lived experience. Whole-brain models provide one way to connect these hypotheses to interacting brain regions and their measured responses to treatment.

This working paper reviews existing depression and related data resources. It did not identify an available shared resource demonstrating the complete proposed combination of individual anatomy, structural connectivity, functional imaging, EEG, and clinical outcomes before and after treatment in thousands of patients. This is a finding from a targeted review, not proof that no suitable data exist. No restricted participant-level files were audited. Existing cohorts provide substantial components, and earlier whole-brain modeling studies establish relevant precedent. FNIH MAP-D is a potential collaborator; partnership and access remain to be agreed.

The first investment should support a coordinated feasibility phase: audit existing cohorts with their custodians, agree a tolerable common measurement protocol, obtain complete usable treatment trajectories, and compare model predictions with simpler alternatives. Recruitment should expand after the pilot establishes data yield, costs, and an adequate statistical design. The illustrative target of 3,000 evaluable participants is a planning objective requiring power and feasibility analysis.

The first scientific test is prospective prediction. Lock model predictions before outcomes are known, evaluate brain changes and clinical outcomes in independent participants and sites, and compare performance with clinical history and simpler statistical methods. Predicting response under a received treatment must be distinguished from demonstrating which treatment should be selected. The paper contributes a resource inventory and a program design; it reports no new patient results or validated treatment-selection system.