Download presentation
Presentation is loading. Please wait.
Published byΧάρων Λιάπης Modified over 5 years ago
1
Big data to smart data in Alzheimer's disease: The brain health modeling initiative to foster actionable knowledge Hugo Geerts, Penny A. Dacks, Viswanath Devanarayan, Magali Haas, Zaven S. Khachaturian, Mark Forrest Gordon, Stuart Maudsley, Klaus Romero, Diane Stephenson Alzheimer's & Dementia: The Journal of the Alzheimer's Association Volume 12, Issue 9, Pages (September 2016) DOI: /j.jalz Copyright © 2016 The Authors Terms and Conditions
2
Fig. 1 Illustration of different pathological processes (by no means exhaustive) occurring in the AD brain and how they putatively interact to lead to the same clinical phenotype. Such drawings are often used by scientists to “formalize” their hypotheses and identify the relationship between different processes. It integrates information from big data studies (e.g., GWAS) with insights about the underlying neurobiology. The purpose of mechanism-based modeling and simulation is to bring these relationships to life by simulating time-dependent and concentration-dependent changes based on equations that describe the specific biochemical processes and ultimately constrained by a number of clinical phenotypes (right side). For instance, synaptic activity dependent formation of beta-amyloid peptides can be simulated (see text) using data constrained by human SILK studies, appropriate enzyme properties, and the forward and backward rate constants of peptide oligomerization. Such models could, in principle, also simulate in a quantitative fashion the impact of therapeutic interventions at specific points in the diagram and therefore support drug discovery and development programs. Alzheimer's & Dementia: The Journal of the Alzheimer's Association , DOI: ( /j.jalz ) Copyright © 2016 The Authors Terms and Conditions
3
Fig. 2 Illustration of the multimodal processes that describe the complexity of going from a single gene (in this case, the huntingtin gene) to fully understand the pathology that leads to the multiple clinical phenotypes in Huntington's disease patients. The increasing complexity when going from one level to the next necessitates the introduction of advanced mathematical modeling and simulation approaches that fully embraces nonlinear and stochastic descriptions of the neurophysiological processes that ultimately leads to clinical phenotypes. For example, although the basic driver of the pathology is the mutated huntingtin gene, its effects on behavior are related in complex nonlinear ways to other processes to the point that is not clear what the optimal target modulation approach would be. In addition, environmental factors or other genotypes likely affect the relative contribution of these pathologic processes to the clinical phenotype. Alzheimer's & Dementia: The Journal of the Alzheimer's Association , DOI: ( /j.jalz ) Copyright © 2016 The Authors Terms and Conditions
4
Fig. 3 Steps for building predictive models. Starting from integrated databases, causal relationships can be identified using not only statistical analysis but also approaches where domain expertise is formalized. These relationships can be tested in biological experiments, together with clinical neuroimaging and neuropathology data and quantitative complex computer models can be developed. Parameters of this model are constrained by clinical data, and predictions can then be tested against actual clinical outcomes. We anticipate a series of interactive steps that will ultimately result in more complex and predictive models. Alzheimer's & Dementia: The Journal of the Alzheimer's Association , DOI: ( /j.jalz ) Copyright © 2016 The Authors Terms and Conditions
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.