Building a Personalized Virtual Brain with Alzheimer’s to Guide Clinical Decisions
The brain is a complicated system whose different parts interact to support a variety of cognitive functions. This complexity makes it difficult to treat diseases such as Alzheimer’s and Parkinson’s, where many different brain areas can be affected, but lead to very similar deficits, such as memory dysfunction. Our research provides a framework of tools to “reconstruct” the brain and build models of different dementias to characterize the unique features of each disease and the final common paths to cognitive impairment. As our work progresses, it will be used to evaluate the potential of therapeutic interventions to help identify treatment targets, or areas of the brain that, if treated, are most likely to result in the best outcome for the individual.
This research project is leading us towards a personalized medicine approach to understanding, preventing and treating brain disorders, specifically Alzheimer’s disease (AD) and Parkinson’s disease (PD), using a network dynamics approach via TheVirtualBrain.
With this grant from BrightFocus Foundation, we are characterizing commonalities and differences in brain network dynamics across AD & PD and mapping the underlying biophysical substrates to individual clinical profiles using TheVirtualBrain (TVB). TVB is a robust brain modeling platform that allows us to simulate network dynamics safely in a virtual environment, in contrast to traditional clinical trials or direct patient testing. The first aim directly tests the hypothesis that AD and PD can be differentiated by specific spatiotemporal patterns of local and distributed processes in the brain. The second aim tests the hypothesis that disease-specific alterations in brain network communication are detectable in prodromal forms of AD and PD and can be used to prevent disease progression and predict clinical outcome. The third aim identifies the underlying common and disease-specific biophysical substrates that lead to alterations in large-scale network dynamics. Importantly, here we identify the biophysical substrates that best predict an individual’s disease trajectory and clinical outcome. TVB is used to evaluate the clinical potential of therapeutic interventions early in development, thus helping to ensure that such efforts converge on targets that are most likely to have the best outcome.
Intensive efforts are underway to build large empirical neuroimaging datasets specific to AD and PD, yet we lack the framework to link these data with the brain function of individual patients. TheVirtualBrain addresses that need by providing a computational and theoretical framework for simulating whole-brain networks. We will use structural and functional patient data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), the Sydney Memory and Ageing Study (Sydney MAS), and the Parkinson’s Progression Markers Initiative (PPMI) to set the initial constraints for estimation of the model parameters.
The impact of our work has both direct and indirect effects on alleviation of human suffering, particularly for those afflicted with dementia. In this research, TVB acts as a “computational microscope” that allows for the inference of internal states and processes that are otherwise invisible to brain imaging devices. TVB is used to evaluate which biophysical model parameters best express the network alterations in AD and PD. Our project provides the basis for more deliberate integration of computational neuroscience and clinical approaches for diagnosis and treatment of brain disorders. The implications for developing targeted therapeutics are clear, where computational models based on a patient’s own data help to guide diagnoses and inform the choices of individualized interventions for the best chance of success. Using TVB as a means to characterize the biophysical parameters that differentiate dementia sub-types imparts great promise for improved early diagnosis and prognosis, as well as treatment success.
About the Researcher
Dr. Randy McIntosh did his undergraduate work at the University of Calgary, graduating in 1987, and then remained there for the next 2 years to complete his masters in psychology. Next he went to the University of Texas at Austin, where he obtained his PhD by 1992 and afterwards began his career as a neuroscientist at the National Institute of Health (NIH). Now, Dr. McIntosh conducts his work at the Rotman Research Institute (RRI) at Baycrest Health Sciences, where he has been a scientist for 22 years. In addition he has served as the RRI’s director since 2008, as Baycrest’s vice president of research from 2009-2017, and is a professor in the Department of Psychology at the University of Toronto. At RRI, his research emphasizes the network operations of the brain and involves parallel development of theory and methods. The theoretical perspective puts forth the notion of neural context, wherein the contribution of a given brain area plays to a process is constrained by other interacting regions. In contrast, the methodological advances have emphasized statistical methods that capture the multivariate nature of brain function. Dr. McIntosh pioneered the application of ‘structural equation modeling’ to neuroscience as a way to measure the effect brain regions have on each other (effective connectivity). Additionally, he developed an extension of ‘partial least squares’ as a method to relate the spatiotemporal patterns of brain activity to behavior, experimental manipulations, or group characterization. His most recent advancement is a convergence of theory and methods in the development of the TheVirtualBrain (TVB, thevirtualbrain.org), which is the culmination of an 11-year international collaboration. TVB is large-scale neural modeling platform that directly uses neuroimaging data to parameterize a model. Because individual data can be used, any person’s brain can become ‘The Virtual Brain.’
First published on: July 6, 2017
Last modified on: July 13, 2020