Usefulness of data from magnetic resonance imaging to improve prediction of dementia: population based cohort study

摘要

STUDY QUESTION Do data from magnetic resonance imaging (MRI) improve the prediction of dementia compared with conventional risk factors in a population based setting? SUMMARY ANSWER There were no significant differences in the discrimination performance of the conventional risk model when compared with models incorporating MRI data, though some improvements in accuracy of classification were observed. WHAT iS KNOWN AND WHATTHIS PAPER ADDS Dementia risk prediction has conventionally been based on sociodemographic, neuropsychological, health, lifestyle, physical function, and genetic variables. Addition of MRI variables, including white matter lesion, brain, and hippocampal volumes, to a conventional risk factor model did not result in significant improvement in discrimination for incident dementia. Additional analyses showed improvement in some prediction markers, such as reclassification and prognostic separation, that could be useful in some settings.