The brain is tuned to operate near a critical phase transition between disordered and ordered states by control parameters such as the balance between functionally excitatory and inhibitory neurotransmission. Deviations from this balance are linked to multiple brain disorders making the emergent dynamics a potential mechanistic biomarker. Indeed, neurophysiological biomarkers appear promising for applications in precision medicine for brain disorders.
However, in complex systems, localization of such biomarkers is confounded by the fact that near criticality, emergent long-range correlations and propagation of signals leads to any nodally or edge-wise localized ‘mechanism’ to both influences and be influenced by other components of the system. In observational data, this kind of “activity mixing” will cause the impact of any localized mechanism or dysfunction to be diffused across a wider network. .
We advance here the idea that fitting of generative models whole-brain dynamics with individual observational data could provide an inverse solution to activity mixing. We used a multi-objective gradient optimization approach to fit local and global coupling parameters of the Hierarchical Kuramoto model, aiming to reproduce dynamics similar to those in real data in terms of both functional connectivity and criticality. Using depression as a model system, we found that the fitted model parameters were more strongly correlated with transdiagnostic depression symptom scores than the underlying observational neuroimaging data. These findings serve as a proof-of-concept that personalized whole-brain modeling can enable reverse engineering of mechanistic biomarkers from observational data.
Contact: Alina Suleimanova, alina.suleimanova@aalto.fi
Aalto University, Helsinki
Additional Authors:
Alina Suleimanova 1, Vladislav Myrov 1, Samanta Knapič 1,2, Paula Partanen 2, Wenya Liu 1,2, Maria Vesterinen 2, Satu Palva 2,3, J. Matias Palva 1,2,3
1: Department of Neuroscience and Biomedical Engineering, Aalto University, Espoo, Finland 2: Neuroscience Center, Helsinki Institute of Life Science, University of Helsinki, Helsinki, Finland 3: Centre for Cognitive Neuroimaging, School of Neuroscience and Psychology, University of Glasgow, Glasgow, UK