Multiscale Predictions Of Mechanical Response And Computational Circuit Dynamics After Traumatic Brain Injury

Traumatic brain injury (TBI) is a major health concern that impacts millions annually. Experimental modeling of TBI has shown many cellular mechanisms of injury, but the connections between cellular dysfunction and microcircuit deficits are poorly understood. Recent developments in computational network models can help connect structural and functional deficits in injured circuits to specific cellular pathologies. Here, we develop a method for rapidly translating induced mechanical strains after impact to regions of brain tissue and investigate how cellular dysfunction due to strain alters the dynamics of local circuitry. We utilized planar spring-mass-damper network models with machine learning adaptation to translate kinematic loading to regional strain prediction for simulated sinusoidal pulses, American football helmet linear impactor testing, and National Football League impact reconstructions. Our approach predicted regional strains with lower computational cost than currently available finite element methods and can rapidly provide strain estimates for local circuits for use in time-sensitive clinical situations. We next investigated how known cellular dysfunctions following regional strains could propagate and alter local network dynamics. Building on previous studies using biological spiking computational network models, we show decreased oscillatory behavior within local networks after neurodegeneration and assess the roles of activity-based neuron phenotypes in oscillation development. Finally, we investigated circuit implications of mechanically sensitive GluN2B containing N-methyl-D-aspartate receptors (NMDAR) dysfunction after mild injury. We show that the increased ion flux through NMDAR immediately increases injured neuron activity rate and propagates to downstream connected circuitry. Additionally, we develop the role of NMDAR dysfunction in impaired recall of learned patterns within injured networks and demonstrate lesser deficits in both pattern acquisition and retention. Together, we developed computational models to predict the mechanical response and circuit dynamics after traumatic brain injury.