We are offering an opportunity for a Short-term Scientific Mission (STSM) with Dr Catherine Drysdale (Lancaster University). The focus of the mission will be on “A Demographic Kernel Framework for Predicting Psychosis Risk Across Urban Space.” For more information, please see below. GREATLEAP can provide up to €2,500 to support the STSM.
To apply, please email your motivation (1 page) and short CV (2 pages) to c.drysdale@lancaster.ac.uk before 4 March 2026.
—————–
Location: Lancaster University (Bailrigg, Lancaster, LA1 4YW, UK), Department of Mathematics and Statistics (MARS – Mathematics for AI in Real-world Systems)
Host: Dr Catherine Drysdale (Lancaster University)
Possible period and length of visit: Autumn 2025, for up to 2 months, full-time placement.
Short description of the proposed project (200–500 words):
This project builds on the findings of our previous seedcorn-funded work (Round 4 – University of Birmingham – Place-Based Early Detection and Prediction Modelling for Psychosis), which developed a geospatial risk model for First Episode Psychosis using ten years of clinical data from an urban region in England. Owing to the marked spatial heterogeneity and temporal instability in psychosis incidence, static hotspot detection methods are insufficient for early detection. To address these challenges, more robust spatial similarity structures are needed to support reliable hotspot detection rather than prediction. To that end, developed a new distance-based matrix that encoded similarity between neighbourhoods using their historical First Episode Psychosis counts, incorporating both spatial proximity and temporal incidence patterns. Applying this matrix within a hotspot-oriented clustering framework, we identified persistent psychosis risk clusters that successfully captured 70–75 percent of high-incidence neighbourhoods in held-out years. This demonstrated both the feasibility and the scientific value of constructing bespoke distance structures based on past burden, providing a stronger mathematical foundation for geospatial hotspot detection.
The proposed GREATLEAP project will advance this work by constructing new spatial distance measures for neighbourhood-level psychosis risk analysis. Rather than relying solely on geographic proximity and historical incidence patterns, we will generate distance matrices that incorporate ethnicity, deprivation, crime exposure, service utilisation. By embedding these demographic and social determinants into the definition of spatial similarity, the project will produce a richer and more meaningful representation of how neighbourhoods relate to one another in terms of underlying risk structure.
The visiting researcher will work collaboratively within mathematically-led-research setting at Lancaster, pairing their demographic insight with the host’s mathematical expertise to support the interpretation and construction of population-level similarity structures. This collaboration will enable rigorous integration of demographic data with advanced mathematical modelling. Applicants from any career stage are welcome, provided they bring a strong interest in demographic or population-level analysis and an ability to grasp mathematical concepts.
By explicitly combining mathematical modelling with demographic insight, the project supports GREATLEAP objectives by creating interdisciplinary capability, generating transferable methodology, and producing high-quality preliminary results for national-scale prediction models. It also builds computational and conceptual bridges between population health, spatial epidemiology, and modern kernel methods.
Possible outputs / outcomes
A publishable article submitted to an appropriate journal such as Health and Place or Spatial and Spatio-temporal Epidemiology.
A reproducible computational pipeline for kernel-based geospatial modelling.
A generalisable mathematical framework for demographic similarity kernels.
Foundations for a subsequent grant application in mental health hotspot depiction