Applying computational modelling to clinical neuroscience

09:05-09:30 Personalised clinical decisions based on computational brain model predictions: What do we need and how can we get it?

Dr Petra Ritter, The Charité Universitätsmedizin Berlin, Germany

Abstract

Dr. Ritter will speak about The Virtual Brain, a framework for computational modelling of full-brain activity that can be related to the data generated in typical neuroimaging protocols. The key innovation is that an individual’s data on brain structure is used to set the initial constraints for model parameters, such as network connection strength and conduction rates. This has profound implications for the understanding and treatment of brain processes and disorders such as those occurring in aging. Extensive simulations on supercomputers allow building a comprehensive library of dynamical regimes that systematically catalogues functional brain modes and underlying candidate mechanisms of neuronal interactions and their age or disease related changes. Such a comprehensive data base is prerequisite for making The Virtual Brain a useful clinical tool.

• Listen to the audio recording

09:30-10:00 Translational neuromodeling

Professor Klaas Enno Stephan, University of Zurich and ETH Zurich, Switzerland

Abstract

For many brain diseases, particularly in psychiatry, we lack objective diagnostic tests and cannot predict optimal treatment for individual patients. This presentation outlines a translational neuromodeling framework which aims at establishing “computational assays” for inferring subject-specific mechanisms of brain disease from non-invasive measures of behaviour and neuronal activity.  Such assays, derived from a generative modelling approach, may provide a formal basis for differential diagnosis and treatment predictions in individual patients and, eventually, facilitate the construction of pathophysiologically grounded disease classifications. The framework presented emphasises the importance of prospective validation studies in patients and of concrete clinical problems for providing benchmarks for model validation. The presentation will show some early (and mostly very simple) proof of concept studies in patients, and outline the opportunities and challenges that lie ahead.

• Listen to the audio recording

10:00-10:30 Discussion

10:30-11:00 Coffee

11:00-11:30 Channelopathies: linking ions to cognition through Dynamic Causal Models

Dr Rosalyn Moran, Virginia Tech, USA

• Listen to the audio recording

11:30-12:00 From differential diagnosis to cognitive restoration in Parkinson’s disease and dementia

Professor James Rowe, University of Cambridge, UK

Abstract

Better treatments for neurodegenerative disorders require a clear mechanistic understanding of the cascade of effects from cell biology through (dys)functional neural networks to complex behavioural syndromes. Computational modelling is critical to this endeavour. I will use examples from several disorders, including Parkinson’s disease, progressive supranuclear palsy and frontotemporal dementia to illustrate our approach. I will argue for a dimensional approach to cognition and behaviour that cuts across traditional diagnostic boundaries, building on the RDoC framework for mental health disorders, and illustrating this with impulsivity and disorders of response inhibition. I will show how formalising action and inhibition as accumulation-to-threshold decisions can provide insights into disease that are not apparent with traditional cognitive outcomes like mean RT and accuracy. I will then examine a network for response inhibition, and its abnormalities in Parkinson’s disease and Frontotemporal dementia. Model-based and data-driven analysis of functional brain networks will be compared, including dynamic causal models for fMRI and MEG. I will present the results of randomised placebo-controlled double-blind crossover functional imaging studies that assess the restoration of network function by noradrenergic and serotonergic reuptake inhibition in clinical disorders associated with noradrenaline and serotonin deficiency respectively. Finally I will touch on models to predict treatment response in highly heterogeneous disorders. Together, these studies confirm that computational models with patient data can facilitate the development of novel therapeutic interventions and individualized therapy while validating preclinical models of brain function and disease.

• Listen to the audio recording

12:00-12:30 Discussion

13:30-14:00 Computational nosology and precision psychiatry

Professor Karl Friston FMedSci FRS, University College London, UK

Abstract

This presentation provides an illustrative treatment of psychiatric morbidity that offers an alternative to the standard nosological model in psychiatry. It considers what would happen if we treated diagnostic categories not as putative causes of signs and symptoms – but as diagnostic consequences of psychopathology and pathophysiology. This reconstitution (of the standard model) opens the door to a more natural formulation of how patients present – and their likely response to therapeutic interventions. In brief, we describe a model that generates symptoms, signs and diagnostic outcomes from latent psychopathological states. In turn, psychopathology is caused by pathophysiological processes that are perturbed by (aetiological) causes; such as predisposing factors, life events and therapeutic interventions. The key advantages of this nosological formulation include: (i) the formal integration of diagnostic (e.g. DSM) categories and latent psychopathological constructs (e.g., the dimensions of RDoC); (ii) the provision of a hypothesis or model space that accommodates formal evidence-based hypothesis testing or model selection (using Bayesian model comparison); (iii) the ability to predict therapeutic responses (using a posterior predictive density), as in precision medicine and (v) a framework that allows one to test hypotheses about the interactions between pharmacological and psychotherapeutic interventions. These and other advantages are largely promissory at present: the purpose of this talk is to suggest what might be possible, through the use of idealised simulations. These simulations can be regarded as a (conceptual) prospectus that motivates a computational nosology for psychiatry.

• Listen to the audio recording

14:00-14:30 Brain network disturbances in mood disorders

Professor Michael Breakspear, QIMR Berghofer Medical Research Institute, Australia

Abstract

Mood serves to adjust our internal somatic state to match the expectations we hold about our social milieu. This requires a dynamic interplay between cognition and interoception. Mood disorders – major depression and bipolar disorder – are characterised by episodic disturbances in arousal, appraisal and socialization. In this talk, evidence will be provided that mood disorders are associated with structural, functional and effective disturbances in brain networks that integrate interoception, social appraisal and cognitive control. However, unlike schizophrenia, the structural “core” of the brain is preserved. These findings position mood disorders as disturbances in the precision with which we hold interoceptive beliefs, and the degree to which we adjust our social expectations following social surprise. Networks supporting broader cognitive and perceptual dynamics are relatively preserved.

• Listen to the audio recording

14:30-15:00 Discussion

15:00-15:30 Tea

15:30-16:00 Predicting outcome and seizure freedom in drug-resistant epilepsy

Dr Viktor Jirsa, University Aix-Marseille, France

Abstract

Starting from first principles of the theory of slow-fast systems in nonlinear dynamics, we conceptualize seizure dynamics mathematically and establish a taxonomy of seizures based on seizure onset and offset bifurcations. It will be demonstrated that only five state variables linked by integral-differential equations are sufficient to describe the onset, time course and offset of ictal-like discharges as well as their recurrence. These state variables define the model system called the Epileptor, where two state variables are responsible for generating rapid discharges (fast time scale), two for spike and wave events (intermediate time scale) and one permittivity variable (slow time scale). The permittivity variable captures effects evolving on slow timescales, including extracellular ionic concentrations and energy metabolism, with time delays of up to seconds as observed clinically.

Extending this generic approach towards human brain networks, personalized connectivity matrices of human epileptic patients using Diffusion Tensor weighted Imaging (DTI) will be reconstructed. Subsets of brain regions generating seizures in patients with refractory partial epilepsy are referred to as the epileptogenic zone (EZ). During a seizure, paroxysmal activity is not restricted to the EZ, but may recruit other brain regions and propagate activity through large brain networks, which comprise brain regions that are not necessarily epileptogenic. The identification of the EZ is crucial for candidates for neurosurgery and requires unambiguous criteria that evaluate the degree of epileptogenicity of brain regions. Stability analyses of propagating waves provide a set of indices quantifying the degree of epileptogenicity and predict conditions, under which seizures propagate to nonepileptogenic brain regions, explaining the responses to intracerebral electric stimulation in epileptogenic and nonepileptogenic areas. The predictive value of our seizure propagation model will be demonstrated by validating it against empirical patient data. In conjunction, our results provide guidance in the presurgical evaluation of epileptogenicity based on electrographic signatures in intracerebral electroencephalograms.

• Listen to the audio recording

16:00-16:30 Brain network dynamics in epilepsy: diagnosis, prognosis and heritability

Professor Mark Richardson, King’s College London, UK

• Listen to the audio recording

16:30-17:00 Discussion

09:00-09:30 Network control theory offers a fundamental mechanism of executive function

Dr Danielle Bassett, University of Pennsylvania, USA

• Listen to the audio recording

09:30-10:00 Testing proposed mesocircuit mechanisms underlying recovery of conciousness with analysis of local and global dynamics of the EEG

Dr Nicholas Schiff, Cornell University, USA

Abstract

This presentation will discuss the “mesocircuit hypothesis” for the key role of the human anterior forebrain mesocircuit in recovery from disorders of consciousness following multi-focal brain injuries. The mesocircuit model makes several specific predictions for the co-variation of measures reflecting progressive restoration of cellular and circuit-level functional integrity following severe deafferentation produced by brain injuries. The talk will focus primarily on the testing of model predictions using quantitative measurements of local and global dynamics of the human electroencephalogram (EEG). Related predictions and measurements from neuroimaging studies of patients with disorders of consciousness and experimental animal studies will also be discussed. Validation of the EEG predictions derived from this large-scale brain network model may provide a set of measures useful to track recovery and predict the impact of different therapeutic inventions in the injured brain on an individual basis.

• Listen to the audio recording

10:00-10:15 Discussion

10:15-10:30 Coffee

10:30-11:00 The effect of Deep Brain Stimulation in whole brain activity: a computational perspective

Professor Gustavo Deco, Universitat Pompeu Fabra, Spain

Abstract

Deep brain stimulation (DBS) for Parkinson’s disease (PD) is a highly effective treatment in controlling otherwise debilitating symptoms yet the underlying brain mechanisms are currently not well understood. Whole-brain computational modelling was used to disclose the effects of DBS ON and OFF during collection of resting state fMRI in ten PD patients. Specifically, the local and global impact of DBS in creating asynchronous, stable or critical oscillatory conditions using a supercritical Hopf bifurcation model were explored. It was found that DBS shifts the global brain dynamics of patients nearer to that of healthy people by significantly changing the bifurcation parameters in very specific local brain regions. Further, higher communicability and coherence brain measures during DBS ON compared to DBS OFF were found. These results offer important insights into the underlying effects of DBS as well as in finding novel stimulation targets, which may in time offer a route to more efficacious treatments.

• Listen to the audio recording

11:00-11:30 Reconstruction and simulation of neocortical microcircuitry

Professor Felix Schurmann, Ecole Polytechnique Fédér ale de Lausanne, Switzerland

Abstract

In a recent publication, the Blue Brain Project has presented its data-driven modeling strategy on the example of the neocortical microcircuit of a young rat. An intricate pipeline of tools has been developed, allowing the integration of multi-modal data and the resolution of constraints using algorithms. Using this pipeline, it is possible to go from sparse data to a model that can be systematically validated against data across multiple levels. A wide range of validation data lays the foundation for the model to generalize to interesting questions that have not been put into the model reconstruction process, thus allowing research on the inverse problem.

• Listen to the audio recording

11:30-12:00 Discussion

13:00-13:30 Mapping causal functional contributions in the brain from the game-theoretical analysis of stroke lesions

Professor Claus Hilgetag, University of Hamburg, Germany

Abstract

Strokes affect multiple brain regions and produce multiple functional deficits, acting as a natural experiment in brain perturbation. However, the interpretation of such perturbation data and the inference of regional functional contributions to brain function is made difficult by the multivariate interactions between different brain regions. To address this problem, we used a multivariate, game-theoretical approach to objectively quantify the causal functional contributions of brain regions, and applied it to animal model data, ground-truth simulations, as well as clinical data of stroke lesions and corresponding functional deficits. The approach reliably inferred regional contributions to brain function and revealed a wide quantitative range of contributions of multiple regions to multiple functions. It also indicated functional synergies and redundancies among brain regions and helped to identify potential targets for clinical rehabilitation

• Listen to the audio recording

13:30-14:00 Using The Virtual Brain as an 'informatics microscope' to uncover biophysical parameters of stroke recovery

Dr Ana Solodkin, UC Irvine School of Medicine, USA

Abstract

An exciting advance in the field of neuroimaging is the acquisition and processing of very large data sets (so called “big data”), permitting large-scale inferences that foster a greater understanding of brain function in health and disease. Yet what we are clearly lacking are quantitative integrative tools to translate this understanding to the individual level to lay the basis for personalized medicine.

This challenge will be addressed through a new neuroInformatics modelling platform that has the capacity to track brain network function at different levels of inquiry, from microscopic to macroscopic and from the localized to the distributed. The multi-scale approach, The Virtual Brain (TVB), can effectively model individualized brain activity, provides unique biological interpretable data in stroke.

The talk will show how our results establish the basis for a deliberate integration of computational biology and neuroscience into clinical approaches for elucidating cellular mechanisms of disease including on the individual level via personalized therapeutic interventions. 

• Listen to the audio recording

14:00-14:30 Discussion

14:30-14:45 Tea

14:45-15:15 Stroke dys-connectome: behaviour, structure and function

Dr Maurizio Corbetta, Washington University, USA

Abstract

Since the early days of neuroscience the relative merit of structural vs. functional network accounts in explaining neurological deficits has been intensely debated.  Using a large stroke cohort and a machine learning approach, we show that visual memory, and verbal memory deficits are better predicted by functional connectivity than by lesion location, while visual and motor deficits are better predicted by lesion location than functional connectivity.  In addition, we show that disruption to a subset of cortical areas predicts general cognitive deficit (spanning multiple behavior domains). In a separate set of computational studies we show that these deficits affects the integration/segregation of brain regions. These results shed light on the complementary value of structural vs. functional accounts of stroke, and provide a physiological mechanism for general multi-domain deficits seen after stroke.

• Listen to the audio recording

15:15-15:45 Predicting outcome and recovery after stroke

Professor Cathy Price, University College London, UK

Abstract

Predicting outcome and recovery after stroke is notoriously difficult because the consequences of seemingly similar brain damage are inconsistent across patients. Clearly factors other than lesion site influence the level of recovery. To investigate the main causes of variability, machine learning on a wide variety of data acquired from large cohorts of stroke survivors has been used. This indicates the importance of lesion site, lesion size and years post stroke as the most informative variables for recovery. Although machine learning does not indicate how the predictions can be improved, it does place constraints on the variables that need to be controlled and considered in models of recovery. Hence, this talk will show how predicting outcome after brain damage can be optimised by a combination of model-based and data-led approaches.

• Listen to the audio recording

15:45-16:00 Discussion

16:00-17:00 Panel discussion/Overview (future directions)