Theme: Technology

Project Description

Psychotic disorders are biologically heterogeneous, but symptom-based diagnoses provide limited insight into the neural processes underlying individual differences and treatment response. MRI and PET are expensive and difficult to deploy at scale. Electroencephalography (EEG) offers a direct, temporally precise and comparatively inexpensive measure of brain function, but conventional research protocols require specialist equipment, numerous electrodes and lengthy recordings.

This project will combine large-scale data analysis with device development to create and pilot a low-cost EEG platform that captures clinically relevant neural processes in approximately five minutes. Its central question is how much meaningful information from high-density EEG can be retained when electrode numbers and recording durations are substantially reduced.

The first work package will analyse multisite Bipolar-Schizophrenia Network on Intermediate Phenotypes (B-SNIP) datasets, deriving features spanning resting activity, periodic and aperiodic spectra, auditory evoked responses, time-frequency measures and connectivity. The student will examine associations with psychosis, cognition, pharmacological exposure and neurobiological heterogeneity.

The second work package will emulate low-density, abbreviated recordings within these datasets. Nested cross-validation and leave-site-out validation will identify electrode configurations, features and durations that preserve biologically relevant information, including robustness to missing electrodes and alternative referencing strategies. Pilot analyses suggest substantial information survives marked reductions in recording requirements. These findings and lived-experience input will guide refinement of an existing bespoke EEG device.

The third work package will prospectively evaluate and iteratively refine the device in people with psychosis, assessing signal quality, feasibility, tolerability, test-retest reliability and concordance with research-grade recordings. It will test whether signatures derived from legacy datasets transfer across cohorts, acquisition systems and clinical environments.

The project will establish the spatial and temporal requirements for scalable electrophysiological phenotyping and deliver a practical clinical research platform. The student will gain training in EEG analysis, machine learning, psychosis neuroscience, device and protocol development and prospective translational research.