Jorge Mejias

Building multiscale digital twins for brain disorders

Dissecting the mechanistic origins of many brain disorders is a daunting problem in neuroscience, largely because of the complex and multiscale nature of such disorders. Classical computational approaches also fail to provide satisfactory results here, because modeling in detail more than one level of this multiscale architecture is often computationally prohibitive. In this talk, I will present recent and ongoing work from my lab which shows how one may carefully build biophysically-plausible models of small neural circuits and scale them up (using mathematical and machine learning tools) to reach large-scale brain networks. These so called ‘digital brain twins’ are a useful way to systematically link microscopic abnormalities, such as defects in specific synaptic receptors, with the alterations of brain dynamics observed in neuroimaging studies of brain disorders. I will primarily focus on the case of models of the human brain for schizophrenia, later building parallelisms with other conditions and/or species. Finally, I’ll speculate on the role that neuro-AI and neurotechnology may play in future digital brain twin developments.