UCL’s Progression Of Neurodegenerative Disease (POND) group develops new computational models and techniques for learning characteristic patterns of disease progression using large cross-sectional data sets. Our primary focus is on dementias caused by diseases such as Alzheimer’s, but our techniques have wider application to other diseases and developmental processes.
Some background reading on “POND modelling”:
We are considering a broad approach to modelling disease progression, starting with Hubert Fonteijn’s work in NeuroImage 2012 on event-based models, and also exploring the possibility of determining causal links between events. Events constitute biomarker abnormality, which includes image-based biomarkers such as regional atrophy in the brain, as well as biomarkers such as levels of abnormal proteins in cerebrospinal fluid.
All the while, we ensure clinical relevance in the models through collaboration with the Dementia Research Centre at UCL’s Institute of Neurology.
UCL POND started with an EPSRC-funded project (see our projects for more details) and are coordinators of the EuroPOND (a Horizon 2020 project) and E-DADS (EU JPND) consortia.
The global ageing population has placed neurodegenerative diseases among the biggest public health challenges of 21st century healthcare. It is vital to understand this spectrum of diseases on both mechanistic and phenotypic levels to elucidate differences and similarities that can inform diagnosis, prognosis, monitoring, therapy development, and treatment & care decisions.
Our vision in the POND initiative at UCL is to provide new avenues for understanding the complexity of clinical phenotypes of multifactorial neurological diseases. Disentangling this complexity by identifying signatures of each disease is essential for meeting the challenge.
The platform upon which we will build the tools for achieving this vision is data-driven computational-and-statistical modelling, a set of powerful approaches with the ability to provide fine-grained and uniquely holistic pictures of neurological disease progression. Such emerging technologies will underpin support systems for clinical and drug-development applications, specifically by enabling precision medicine through differential diagnosis, patient staging, and personalised prognosis.
Our strategy for achieving impact within our vision requires a balance between model utility and complexity. Model utility is the end-game focus in order to impact disease management across the full spectrum from patients to medical health professionals and drug-development companies. Model complexity is unavoidable due to the nature of the disease signatures we seek, and requires methodological development, which is one of our group’s strengths.
Data-driven models for Progression Of Neurological Disease
Data-driven models for Progression Of Neurological Disease
Neil’s UKRI Future Leaders Fellowship
Danny’s Wellcome Trust Investigator in Science award
A sprint project to break down barriers to AI research on routinely collected healthcare data
Using MRI in late pregnancy to help predict the best mode of birth for an individual mother and baby.
Data-driven disease progression modelling for differential diagnosis of neurodegenerative diseases and their subtypes
Using MRI in late pregnancy to help predict the best mode of birth for an individual mother and baby.
This project uses POND models to analyse autosomal dominant Alzheimer’s disease and investigates the relation between disease progression, biomarker evolution, genetic mutation status and cognitive performance. Key outputs include data-driven sequences of disease progression, subtype genetic profiles, cognitive performance trajectories, and early pre-symptomatic biomarkers.
A list of our past projects
Since 2014 the UCL POND team has convened a biennial workshop/conference meeting focussed on data-driven modelling of disease progression. The first few were part of the EuroPOND consortium that we led.
For more information, see pondmodels.net.