EPAD’s computational models, working on geographically-diverse, high-quality datasets, help identify subgroups in cognitive function evolution
The IMI project EPAD, which sought to better understand dementia risk, has released another trove of data from their three-year dementia risk study. In a paper titled “Disease modelling of cognitive outcomes and biomarkers in the European Prevention of Alzheimer’s Dementia Longitudinal Cohort”, published in the journal Frontiers in Big Data, the project partners have described their work on teasing out the different trajectories of people at risk of developing dementia.
The modelling suggests four distinct trajectories, from class zero, characterised by individuals having the highest levels of cognitive functioning with no signs of impairment at the beginning of the study and no decline throughout the course of the study, to class three, which describes individuals who showed the most obvious signs of early cognitive and/or functional impairment in the beginning of the study and continued to show impairment upon follow-up.
The ultimate aim of this work is to establish whether cognitive tests and biomarkers can be used to figure out who is likely to develop dementia, because early intervention offers the best chance of better outcomes. EPAD’s dataset covers a wide range of measurements, including cognitive outcomes, neuroimaging, cerebrospinal fluid biomarkers, genetics and other clinical and environmental risk factors. This is important because individual data points don’t tell us enough. The more diverse and dynamic the data, the more likely statistical and machine learning will be successful at drawing robust conclusions from the data.
While such longitudinal studies are ongoing around the world, the EPAD study groups were unique in their geographical breadth. This diversity of populations is important because it gives us data that researchers can apply to a wider swathe of the human population than those confined to a specific location.
This latest dataset is being added to the AD Workbench, where other EPAD datasets are already available, making it findable, accessible, interoperable and reusable (according to the FAIR principles) so that other researchers can use it in their own dementia studies. The project also made available a biobank of patient samples collected during the study.
How they did it
The researchers applied state-of-the-art modelling and stratification methods to the data in order to characterise disease progression and biological heterogeneity within the group. They used class-specific mixed effects models (statistical models that contain both fixed effects and random effects) and clustering to work out the different disease trajectories, and explore whether participants can be stratified into homogeneous subgroups according to their cognitive functioning evolution and biomarker profile.