Each week, we will be publishing layman’s abstracts of new articles from our prestigious portfolio of journals in statistics. The aim is to highlight the latest research to a broader audience in an accessible format.

The article featured today is from the Canadian Journal of Statistics, with the full article now available to read in Early View here.

Dai, N., Kang, H., Jones, G.L., Fiecas, M.B. and (2021), A Bayesian latent spatial model for mapping the cortical signature of progression to Alzheimer’s disease. Can J Statistics. https://doi.org/10.1002/cjs.11588

Alzheimer’s disease (AD) is a complex neurodegenerative disease with biological markers that precede cognitive and behavioral symptoms by many years. Imaging biomarkers are commonly used as part of the diagnosis of AD due to its noninvasiveness, availability, and high sensitivity. Indeed, the spatial topography of cortical atrophy, which can be measured using magnetic resonance imaging (MRI), strongly correlates with cognitive deficits, and the degree of atrophy in specific structures of the brain has been a useful diagnostic imaging biomarker for Alzheimer’s disease. MRI data collected longitudinally can characterize the rate of cortical atrophy. In this study, the authors developed a Bayesian latent spatial model that leveraged the spatial topography of the thinning of the cortex to model the time-to-conversion from mild cognitive impairment (MCI) to AD. Their model builds on previous works related to spatial statistics and time-to-event analyses. They analyzed the longitudinal MRI data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), a multisite study whose data and samples are available to researchers worldwide. Using synthetic data, they validated their model, and showed its excellent performance with respect to accuracy and statistical power. Furthermore, they analyzed the longitudinal MRI data from ADNI, and showed that their model yielded larger effect sizes and greater spatial extent of the findings with respect to those reported in the literature. Altogether, their model opens up the possibility of examining the relationship between the temporal and spatial dynamics of imaging data with clinical events. The code for the proposed model is publicly available.