Computational methods for disentangling the effects of mixed pathology in neurodegenerative disease

This project redefines the spectrum of neurodegenerative conditions via a new understanding of mixed pathologies provided by novel computational modelling approaches. Mixed pathology affects most individuals with neurodegenerative diseases and is a key confounder in clinical trials yet is not widely considered or quantified in research studies as several pathologies lack specific biomarkers. This project builds new computational tools that learn the contributions of mixed pathology to non-specific biomarkers such as structural magnetic resonance imaging by accounting for major confounding factors including disease progression and within-pathology subtypes. This will enable individuals to be given a pathology profile describing the different pathologies they have and the quantity of each. These pathology profiles will then be used to improve patient stratification and prediction of clinical outcomes, to understand the links between pathologies and symptoms, to generate hypotheses on treatment strategies, and to shed new light on the shared mechanisms of concurrent pathologies. Whilst the project focusses on neurodegenerative diseases, the aim is to develop tools that will ultimately have broader applications across a wide range of complex long-term health conditions.