Skip to main content
The William Harvey Research Institute - Faculty of Medicine and Dentistry

Breadcrumb

  1. WHRI
  2. Study
  3. Advanced AI for Multi-modal Spatial Biology - PhD Programme
  4. Projects and Partners
  5. BBSRC-DFA_2026_08

AI-Enabled Cross-Modal Integration of Electrophysiology and Spatial Transcriptomics to Decode Sleep Architecture

Code: BBSRC-DFA_2026_08 (CASE)

Primary Supervisor: Dr Zhongbo Chen
Email: zhongbo.chen@qmul.ac.uk 
Institute: Centre for Preventive Neurology

Secondary Supervisor: Dr Lin Wang
Email: lin.wang@qmul.ac.uk 
Institute: School of Electronic Engineering and Materials Science

CASE Partner: AstronauTx

Abstract:

This project will generate computational tools to integrate high resolution spatial transcriptomic data with in-depth electroencephalogram (EEG) recordings to investigate how sleep architecture relates to molecular organisation of the brain.  

Sleep dysregulation increases with age and disruptions in sleep are associated with poor brain health. While these disruptions are well characterized on EEG/polysomnography recordings, the underlying biological mechanisms remain poorly understood. 

By leveraging 10x Visium spatial transcriptomic profiles from brain tissue of human donors alongside matched single-nucleus RNA-sequencing, and detailed EEG data, the candidate will develop cross-modal registration algorithms and scale-aware deep learning models. These computational methods will be applied to aligning EEG-derived features with cell-type proportions and pathway-level gene expression signatures. Emphasis will be placed on interpretable AI, enabling biological attribution of EEG signals to specific cellular processes.  

The main computational challenges include integrating heterogeneous data types with different spatial and temporal resolutions, aligning electrophysiological features with anatomically resolved gene expression domains, and developing interpretable models capable of linking dynamic neural activity to static molecular maps. As such, the candidate will develop multimodal deep learning frameworks, approaches to infer shared latent structures across EEG-derived features and spatial transcriptomic profiles.  

The candidate will have the opportunity to join AstronauTx, a biotechnology company dedicated to improving sleep architecture to gain direct translational and industrial exposure in the AI applications.  

This doctoral project aligns strongly with the Doctoral Focal Award vision in advanced AI for multimodal spatial biology by developing novel, scalable, interpretable AI tools to decode complex spatial and functional datasets. 

Lay Summary:

Sleep is a fundamental biological process that reflects coordinated activity across brain networks. Scientists can measure sleep using electroencephalography (EEG), which records electrical activity from the scalp and reveals characteristic rhythms such as deep sleep and rapid eye movement (REM) sleep. However, while EEG captures fast-changing brain activity, it does not directly show how this activity relates to the underlying cellular and molecular organisation of the brain. 

At the same time, new technologies such as spatial transcriptomics allow researchers to map which genes are active in precise locations within brain tissue. This produces high-resolution maps of cell types and molecular pathways across different cortical layers and regions. These data provide detailed structural and molecular information, but they are static and do not capture dynamic brain function. 

This project focuses on developing advanced computational tools to connect these two complementary data types. Using artificial intelligence (AI) and machine learning, we will build models that link patterns of electrical activity measured during sleep with spatial maps of gene expression. The goal is not simply to predict genes from EEG, but to create biologically interpretable frameworks that align functional brain rhythms with molecular architecture. 

By developing scalable algorithms for cross-modal integration, this research will provide new ways to decode how large-scale brain activity relates to cellular organisation. These tools will be broadly applicable to neuroscience, enabling researchers to integrate functional recordings with spatial molecular maps in both experimental and human studies. 

Aims and Objectives:

Aim 1: Develop AI frameworks for spatially informed EEG representation learning 

Aim 2: Integrate spatial transcriptomics using AI models 

Aim 3: Scale multimodal AI models to large human EEG cohorts  

Back to top