Workshop at CNS 2013, Paris

VALIDATING NEURO-COMPUTATIONAL MODELS OF NEUROLOGICAL AND PSYCHIATRIC DISORDERS

Organiser and chair: Dr. Basabdatta Sen Bhattacharya, University of Lincoln, UK     (BBhattacharya@lincoln.ac.uk)

Co-organiser: Prof. Fahmida Chowdhury, National Science Foundation, USA (fchowdhu@nsf.gov)

Co-Chair: Dr. Rosalyn Moran, Assistant Professor, Virginia Tech Carilion Research Institute, USA (rosalynj@vtv.vt.edu

 http://www.cnsorg.org/cns-2013-workshops-program

Date: 18th July 2013, 9:00 to 17:30 hrs

Recent years have seen a widespread interest in applying computational models to underpin the neural correlates in neurological and psychiatric disorders, which is essential for drug discovery, disease prediction and better diagnostics. Neuro-computational models are abstractions of highly complex biological circuitry and/or phenomena at a level appropriate to the modeller’s target `problem’. An essential condition for models simulating real world phenomena to be ‘usable’ is to validate them in order to avoid erroneous understanding and potentially conflicting predictions. In other words, a model can be deemed useful as a tool to aid the understanding and treatment of disease conditions only if it is validated with experimental data. Currently, there is a rich repertoire of computational models, mimicking the functionalities and behaviour of various brain parts. However, the immense diversity in modelling and validation approaches across the globe makes it difficult to compare results, even for similar brain functionality. In addition, validation techniques as well as the experimental data used for validation are not ‘homogeneous’. Moreover, being a multidisciplinary field, a structured and co-ordinated approach to benchmark and/or set standards for validation methods and techniques is yet to be initiated. The aim of this workshop is to bring together Engineers and Scientists who work on modelling brain behaviour to discuss

(a)    Potential methods of meaningful validation of neuro-computational models with experimental data.

(b)   Ways of benchmarking validation of different modelling approaches towards a given goal (e.g. modelling anomalies in EEG), so that different models may be compared in meaningful ways.

(c)    Potential new collaborations and leverage existing ones to advance the field.

The expected outcome is a report or white paper written by the group (including the speakers and any interested persons from the audience) to present their findings and a few action items.

Plan for the day


Welcome and Introduction        



Epileptic transitions  - model predictions and experimental validation

Dr. Basabdatta Sen-Bhattacharya


Dr. Piotr Suffczynski

9:00 - 9:15



9:15 10:00

 Coffee Break

10:15 10:30

Spatio-temporal dynamics of local cortical circuits and DCM



Dr. Dimitris Pinotsis




10:30 11:15

The cake is a lie - Limits to biological realism in 

(neural population) models

Prof Ingo Bojak

11:30 12:15



Discussion


12:30 13:00


Lunch


13:00 – 14:15


Fit and Evaluation of Cortical models


Dr. Udo Ernst


14:15 – 15:00

Coffee Break

15:15 15:30

Basal ganglia–cortical interactions in Parkinsonian patients

Dr. Andre Marreiros

15:30 – 16:15


Understanding interacting neurotransmitter systems in cortical networks through Dynamic Causal Modelling


Dr. Rosalyn Moran


16:30 – 17:15

Discussion and wrap-up : 17:15 – 17:30


Talk Titles and Abstract

  • Dr. Piotr Suffczynski, Associate Professor, University of Warsaw, Poland    (suffa@fuw.edu.pl )

         Talk TitleEpileptic transitions - model predictions and experimental validation
           Abstract: The essence of epilepsy is that a patient displays (long) periods of normal EEG activity (i.e. non-epileptiform) intermingled occasionally with epileptiform paroxysmal activity. The mechanisms of transition between these two types of activity are not well understood. In order to provide more insight into the dynamics of the neuronal networks leading to seizure generation we developed a computational model of thalamocortical circuits based on relevant patho(physiological) data. The model exhibits bistability, i.e., it features two operational states, ictal and interictal, that co-exist. The transitions between these two states occur according to a Poisson process. The predictions of bistable computational model are compared with experimental results from different types of epilepsy. In a number of long-time recordings epileptic transitions are consistent with the model predictions. In some other data sets alternative routes leading to generation of epileptic seizures are observed. Additional physiological mechanisms subsequently implemented in the model to account for the deviations from the Poisson process are discussed.          

  • Prof. Ingo Bojak, University of Reading, UK
            Talk Title: The cake is a lie - Limits to biological realism in (neural population) models
  • Dr. Rosalyn Moran, Assistant Professor, Virginia Tech Carilion Research Institute, USA
            Talk Title: Understanding Interacting Neurotransmitter Systems in Cortical Networks through Dynamic                                                Causal Modelling
          Abstract: The appropriate integration of signals from functionally specialized brain regions is essential for normal cognition and behaviour, and its absence may underpin the symptoms of a broad range of neurological and psychiatric diseases. In this talk I will propose Dynamic Causal Modelling as a ‘mathematical microscope’ that can provide regional, laminar, neurotransmitter and receptor specific assays of functioning brain networks. 
In particular, neuromodulatory chemicals such as dopamine, acetylcholine and serotonin, organise a diverse set of cognitive processes and are implicated in neurodegenerative diseases including Parkinson’s disease and Alzheimer’s disease. Understanding how they interact with primary neurotransmitters in active brain networks is hence an important prerequisite for understanding pathological onset and progression. In the talk, I will present data and a DCM analysis of Parkinsonian animal recordings that reveal changes in connectivity in motor circuits downstream of dopamine loss. From a healthy human population, I will describe how DCM was used to link behavioural improvement under pro-dopaminergic (levodopa) modulation to changes in AMPA and NMDA mediated signalling in prefrontal regions. I will report pro-cholinergic effects in a analysis of healthy adults that delineates where and how acetylcholine modulates perceptual processing and learning. 
Effectively I aim to demonstrate that it is possible to cross many scales of neuroscientific description using data acquired non‐invasively, linking overt behaviour to its molecular components and providing a window to the machinery of the human mind.
            Talk Title: Spatio-temporal dynamics of local cortical circuits and DCM
  • Dr. Udo Ernst, Institute for Theoretical Physics, University of Bremen, Germany
           Talk Title: Fit and Evaluation of Cortical models

            Abstract: Cortical networks are highly complex, consisting of numerous interacting areas with millions of recurrently coupled neurons. From microscopic knowledge alone, it is as yet not possible to build parameter-free models of such networks. Instead, modeling relies on heavy simplifications, on assumptions on relevant mechanisms, and on phenomenological models for investigating and understanding cortical function. This approach necessitates to link model variables to experimental observables, to find correct parameters for highly nonlinear systems, and to critically evaluate competing models or explanations for the same cognitive phenomenon. Here, I will present some ideas and methods on how to fit and evaluate cortical models against experimental data. In particular, the focus will be on models of early visual computation and spatio-temporal feature integration. Finally, I will discuss how these ideas might be adapted to modeling of neurological disorders, and point out challenges for future methodological development.

  • Dr. Andre Marreiros, Max Planck Institute of Biological Cybernetics, Tubingen, Germany (mar@nbi.ku.dk
           Talk Title: Basal ganglia–cortical interactions in Parkinsonian patients



Comments