|PUBLICATION||UbiComp/ISWC'15 Adjunct Adjunct Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2015|
Recent findings from cognitive neuroscience indicate that individual differences in various cognitive traits such as personality and intelligence are reflected in macroscopic brain structure that can be quantified with magnetic resonance imaging (MRI) techniques. While the initial aim of such research was to identify brain regions where structural variation correlated with cognitive functions, the flurry of correlational findings suggest that MRI scans of the human brain can be used to make predictions about individuals. We have confirmed this idea by constructing highly accurate prediction models for a multitude of cognitive traits using a machine learning technique called elastic net. We are currently implementing those prediction models on a web-based service called Neuroprofile where anyone can upload their brain scan to view the predicted cognitive traits. This service will be provided to make the contents of MRI-based brain health check-ups much richer and more informative.