CFE 2019: Start Registration
View Submission - CMStatistics
Title: Mining the UK Biobank for indicators of brain health Authors:  Thomas Nichols - University of Oxford (United Kingdom) [presenting]
Abstract: The UK Biobank contains detailed records on 1/2 million UK residents aged 40 to 69 years at baseline recruitment and, now 10 years later, MRI brain scanning on 100,000 subjects is underway. While `only' 20,000 MRI datasets are available for our present analyses, it is still the largest such brain imaging study every undertaken. We will review several lines of work that looks to establish the link between variation in brain measures and psychiatric outcomes and risk factors. As the coarsest measure of brain health, we examine predictors of age based solely on brain data, producing a `brain age' estimate and examine how the difference of estimated brain age and (true) chronological age predicts psychiatric outcomes. Unlike other efforts, we pay particular attention to producing well-calibrated confidence statements on these brain age prediction. We then look at highly focused analyses to find patterns of brain structure, function and connectivity that associate with depression. While UK Biobank is a population sample with no particular focus on psychiatric disease, with so many subjects we are able to find 100's with indicators of depression and construct a matched sample. We then use multivariate methods like CCA to find brain variables that associate with depression. Overall, the UK Biobank presents a wealth of opportunities for psychiatric neuroimaging research.