Personalized Multiscale Modeling of Brain Dynamics in Clinical High-Risk States for Psychosis: A First Exploration Using the Psychosis-Risk Outcomes Network Dataset

Schizophrenia is a severe and heterogeneous psychiatric disorder characterized by alterations in thought processes, perception, emotional responsiveness and social behavior. Increasing evidence suggests that large-scale brain network dysfunction associated with schizophrenia spectrum disorders is already present during the clinical high-risk for psychosis (CHR-P) stage, a condition characterized by subthreshold psychotic symptoms and an elevated risk of transition to a full psychotic disorder. Multi-scale brain modeling provides a promising framework to investigate such alterations by integrating structural and functional neuroimaging data with biophysical neural mass models, allowing the assessment of large-scale dynamics and excitation/inhibition profiles associated with high-risk states. The aim of this study was to determine whether CHR-P individuals exhibit altered network dynamics compared to healthy controls using subject-specific simulations. To address this, The Virtual Brain (TVB) modeling platform was employed. By integrating multimodal neuroimaging data from 58 subjects (41 CHR-P, 17 healthy controls - HC), we simulated brain dynamics for each of them. For the simulations, we used the Wong-Wang model that allows for extracting neurophysiological information. Specifically, model parameters reflect long-range connections, local inhibitory or excitatory coupling and auto-excitation of the circuits. The optimal values for these parameters are calculated through an optimization procedure where the simulated functional connectivity (FC) matrix is compared to the empirical one until the best match is reached. We calculated the optimal model parameters for seven canonical resting-state networks and compared them between CHR-P individuals and HCs using a linear mixed-effects approach. Exploratory post-hoc contrasts revealed higher global coupling within the frontoparietal and visual networks, as well as increased local inhibitory coupling within the dorsal attention network in CHR-P individuals compared to HCs. These findings point toward subtle, network-specific modulations in large-scale integration and inhibitory dynamics rather than widespread disruptions of global brain organization. Overall, these results suggest that early psychosis risk may involve localized alterations in global and local coupling within cognitive control and sensory networks, emerging against an otherwise preserved large-scale dynamical architecture. Whole-brain computational modeling may therefore provide a sensitive framework to uncover latent alterations in network dynamics that may not be directly observable from empirical data alone. Further longitudinal studies in CHR-P populations are needed to clarify how deviations in connectivity and excitation/inhibition balance relate to transition to psychosis and treatment response, ultimately contributing to the identification of early, mechanistically informed biomarkers.