Organisation:
Department:
Keywords:
Interests:
My PhD Research: Normative Modelling of Resting-State EEG Across the Lifespan for Application in the Early Diagnosis of Neurodegenerative Disorders, Including Alzheimer’s Disease.
Given the rapidly ageing human population, neurodegenerative disorders are becoming an urgent concern. By the year 2050, the proportion of the world’s population over 60 years is projected to almost double, increasing from 12% to 22% (World Health Organization, 2025), and dementia is expected to impact an estimated 152 million people (Alzheimer’s Disease International, 2018).
Alzheimer’s disease accounts for approximately 80% of all dementia cases (Alzheimer’s Association Report, 2025). At present, there is no known cure for Alzheimer’s disease. Current interventions can slow disease progression (Passeri et al., 2022), however a significant challenge is that Alzheimer’s disease is often only diagnosed late in the disease process (Thakur et al., 2025). Moreover, drug development can be extremely costly (Scott et al., 2014; Sertkaya et al., 2024) and slow (Cummings et al., 2016). Therefore, identifying novel biomarkers is of crucial importance for facilitating the early diagnosis of Alzheimer’s disease and developing new therapeutic interventions.
Research Themes:
Current biomarkers are identified through magnetic resonance imaging, positron emission tomography, and cerebrospinal fluid. Unfortunately, these techniques are invasive, costly, and difficult to access (Chen et al., 2024; Fruehwirt et al., 2019).
Electroencephalography (EEG), however, shows potential as an inexpensive, accessible, and non-invasive alternative. Research has demonstrated that EEG can reliably differentiate between participants with neurodegenerative disorders and healthy controls (Benwell et al., 2020; Flores-Sandoval et al., 2023). However, previous studies have relied on case-control designs which focus on average differences between groups, implicitly overlooking individual differences as statistical noise (de Boer et al., 2024; Marquand et al., 2019). This is problematic, as Alzheimer’s disease is a highly heterogenous condition and relying on average differences may hinder the identification of sensitive and specific biomarkers that could lead to earlier diagnosis and developed interventions (Tabbal et al., 2024).
We are therefore building normative models of resting-state EEG (eyes-open & eyes-closed) allowing for inferences regarding the likelihood or absence of Alzheimer’s disease (and other disorders with abnormal EEG patterns), based on richer biological and social data than case-control studies have allowed for. Normative modelling is a statistical approach that involves charting the typical trajectory of a biological measure (such as EEG) across a population, as a function of specific covariates (for example, age and sex) (Marquand et al., 2019; Rutherford et al., 2023).
We will then test whether our normative models have improved sensitivity for detecting Alzheimer’s disease than traditional case-control designs. Normative modelling therefore preserves the individual-level variability within heterogenous populations and may facilitate early detection of Alzheimer’s disease.
Our normative models will be made open source, to allow for iterative refinement of the models by researchers and clinicians worldwide.
So far, we have collected data from:
- >8,000 healthy controls
- >200 individuals with Alzheimer’s disease
- 23 individuals with frontotemporal dementia
- 46 individuals with multiple sclerosis.
Our normative models will therefore be substantially larger than the minimum recommended healthy control sample size of 1,000; indicating that they may outperform smaller, closed datasets (Rutherford et al., 2022).
Implications:
- Make early diagnosis possible, based on rich biological and social data
- Identify novel biomarkers, useful for the development of new therapeutic interventions
- Identify more at-risk groups, that clinicians could assess at an earlier stage with EEG
- Improve the accuracy and precision of normative modelling by making our models open source
- Advance EEG normative modelling by facilitating research into other disorders
- Encourage further research using EEG given its non-invasiveness, cost-effectiveness and demonstrated ability to differentiate between healthy controls and individuals with Alzheimer’s disease, as well as between disease states
Software Expertise:
Through my doctoral research (supervisors: Dr Christopher Benwell and Dr Christian Keitel), I am gaining extensive experience in coding and computational modelling (MATLAB and Python). I also undertook an optional MSc module (Advanced Quantitative Methods) in the first year of my PhD, where I learned JASP and G*Power.
During my MSc in Applied Neuroscience (supervisor: Dr Tom Gilbertson), I used MATLAB and Python to investigate the relationship between apathy in Parkinson’s disease and explore-exploit decisions. Additionally, I learned how to use R Studio in two of the taught modules of my MSc.
My BSc Honours in Psychology research (supervisor: Dr Christopher O’Donnell) on cognitive decline involved creating an online experiment using OpenSesame and Audacity, and I conducted statistical analysis in SPSS.
Collaborators:
I am delighted to be co-authoring with Mrs Mihaela Lyutskanova, Dr Tom Gilbertson, Professor Douglas Steele, Dr Christian Keitel, and Dr Christopher Benwell.