A computational modelling study investigating electrophysiological biomarkers of Alzheimer's disease
Alzheimer’s disease (AD) is a neurodegenerative disorder, pathologically characterised by loss of synapses, accumulation of neurofibrillary tangles and amyloid beta plaques (Alzheimer’s Association, 2018). While AD is the most common underlying cause of dementia, it is difficult to diagnose the disease and determine its progression.
The disease has a long prodromal stage and clinical symptoms are only seen in the late phase of the disease, which makes researching treatments difficult. To improve diagnosis and prognosis in the preclinical stage, electroencephalography-based event-related potential biomarkers are being developed (Olichney, Yang, Taylor, & Kutas, 2011). AD patients manifest abnormal electrophysiological signals compared to healthy adults when performing the semantic category judgement task, and some of the reported electrophysiological biomarkers of AD include the N400 ERP congruency effect (Olichney et al., 2006, 2008), the N400 and P600 ERP repetition effects (Olichney et al., 2002, 2013), and oscillatory power changes (Mazaheri et al., 2018) .
While the cognitive aspects of these biomarkers have been studied at multiple stages of disease progression, the exact relationship between AD pathology and the abnormal electrophysiology has not yet been discerned (Bhattacharya, Coyle, & Maguire, 2011; Palop & Mucke, 2010; Stefanovski et al., 2019) .
This research attempts to determine the relationship between AD pathology and electrophysiology through the development of a biologically detailed computational model to simulate the semantic category judgment task used in literature . The neuronal characteristics of this model are derived from the spiking Selection over time and space model (sSoTS) which has previously been used in visual search literature (Mavritsaki, Bowman, & Su, 2019; Mavritsaki & Humphreys, 2013). This model encompasses the AMPA, GABA, NMDA, and Ca2+ dependent K+ spike frequency adaptation currents which are known to be modulated by AD pathology.
The connectivity of the model for simulating the semantic judgment task is approximated using the mean field approach, which optimizes utilization of computational resources permitting the exploration of a larger parameter space before detailed simulations at the spiking level.
We intend to develop this model in four steps: first, by developing a model to simulate the N400 congruency effect, followed by updating the model to simulate the N400 and P600 repetition effects, then modulating neuronal characteristics to simulate individual and combined roles of AD pathology and lastly simulating the oscillatory power changes.
This is the first step towards developing a biologically detailed model of progressive deterioration of electrophysiology in AD and determining the relationship between AD pathology and electrophysiology. The research will have immense implications for AD research as it could help improve AD diagnosis and prognosis of patients and also aid efforts in identifying targets for medical intervention and tools for personalized care.
This project is funded through internal funding from Birmingham City University and is in collaboration with University of Birmingham, who have allowed access to their BlueBEAR High Performance Computing system.
