Whole-brain computational modelling for the characterisation of and transition between brain states

Whole-brain computational modelling is an emerging area in the mathematical and computational study of the human brain that merges neural data and connectomics with generative mathematical models of brain dynamics. This project will explore the exciting applications of this novel approach to the study of brain states in an effort to improve understanding of the relationship between cognitive functions and spatiotemporal brain dynamics. A particular focus will be placed on the study of states with various levels of consciousness such as wakefulness, deep sleep, anaesthesia, coma and other diseases of consciousness. The characterisation of brain states will be studied using data science and signal processing techniques to extract latent features in neuroimaging and neuroelectric data, such as fMRI or MEG, from a range of patient populations in varying states of consciousness. DTI and fMRI data can be combined with the probabilistic tractography to obtain the structural and functional connectivities, according to some parcellation of brain regions. Promising lines of research exist in the measurement of causality, complexity, entropy production and statistical irreversibility of neural signals using measures from dynamical systems theory, information theory and statistical mechanics. In addition to the characterisation of brain states from neural data, the project will further aim to build generative whole-brain models using a framework built on networks of coupled oscillators and mean field approximations. Existing mathematical models will be extended to incorporate the temporally changing structure of brain networks as a result of network degeneration, a key feature absent from many current whole-brain models. Another feature of the whole-brain models that will be further developed is the inclusion of neurotransmitter dynamics which are key for modelling the effects of pharmacological interventions. From a mathematical perspective, this project will facilitate the development of novel techniques in neural data science as well as dynamics on networks. Using statistical inference and optimisation, whole brain models are traditionally fitted to both the structural and time-series data but could also be fit to the latent features characterising a brain state once these are determined. Furthermore, in this project we will aim to model the forcing of transitions between states by modelling interventions such as deep brain stimulation, transcranial magnetic stimulation or pharmacological stimulation, such as psychedelic therapy. A computational model of this kind could provide an effective way to perturb the system and compare the effects of a range of treatments in a systematic way that is not feasible in vivo. The aims of this project are to facilitate in-silico experimentation that could provide new insight into the relationship between higher level cognitive abilities and the underlying brain dynamics, as well as to motivate the development of novel treatments and biomarkers for diseases of consciousness with the potential for real patient impact in otherwise difficult patient populations. This work falls under the EPSRC mathematical sciences theme as well as some overlap with the healthcare technologies theme and with a particular relevancy to the EPSRC areas of interest in biological informatics and mathematical biology. This project will be joint between the Mathematical Institute and the Centre for Eudaimonia and Human Flourishing.