Computational Modelling and Inference of Neurodegenerative Disease Propagation

Neurodegenerative disease may be the biggest challenge facing 21st Century healthcare. Delaying Alzheimer’s onset by five years would relieve untold suffering and save £10B annually in the UK. This may come from a) treatments that slow disease (none are yet proven), or b) avoiding risk-associated behaviour. Both require deeper understanding of disease initiation and progression. 

Molecular biology has provided fundamental insights into how abnormal toxic proteins appear and propagate killing brain cells and disrupting cognitive ability. However, conflicting theories on the driving mechanisms remain, with little prospect of resolution from bottom-up molecular investigation.

I propose a complementary top-down viewpoint using machine learning to infer underlying disease processes from macroscopic pathological patterns observed in imaging data from large patient groups. This uniquely reveals the variability among patients supporting personalized prevention and treatment. Ultimately, it brings computational models to the forefront of understanding and managing neurodegeneration, cf. computational fluid-dynamics in weather prediction.