Close
Alzheimer’s disease is the most common form of dementia in the elderly. No treatment has been proven to have a disease-modifying effect. One key confound in identifying effective treatments is the heterogeneity of disease onset and clinical presentation, which vary widely. Most likely, different treatment strategies are appropriate for different patients. However finding subgroups of patients with similar treatment needs is also challenging, because they change dramatically over the multiple decades it takes for the disease to run its course.
This project will use disease progression modelling and clustering (unsupervised learning) to identify groups of patients with particular trajectories of cognitive changes during disease progression. It will adapt recent advances in image-based modelling (the SuStaIn algorithm [Young Nat. Comms. 2018] - see figure) to work with cognitive data and utilise large data sets of cognitive test scores and associated outcomes available from historical visits to local memory clinics. The project will also consider the role and implications of this emerging AI technology in the clinic with a view to prototyping medical software for widespread translation and usage. We will also consider linkage of cognitive data to other data types, such as imaging, genetics, and lifestyle factors to help refine the landscape of subtypes and relate them to underlying pathological mechanisms.
Daniel Alexander, Dept of Computer Science