Computational psychiatry: Advancing predictive modeling of neurodegeneration with neuroimaging of Huntington's disease

Abstract

Computational psychiatry is a relatively new field that integrates computational models, biological knowledge, and clinical data to provide much needed predictive power to our understanding of brain disorders. The enterprise is far from straightforward, and difficulties include the definition of psychiatric illnesses, variability from subjective elements in measurement, lack of large systematic databases, and mastering subtle technical aspects behind different measurements. In this perspective article, we make the case for studying Huntington's disease as a model system of neurodegeneration that is largely exempt from many of these limitations. The combination of its well-defined genetic basis and the availability of large neuroimaging databases makes it an outstanding candidate for marrying computational and clinical approaches. We highlight particularly promising directions and sketch a roadmap for a research program, with a focus on producing clinical insights based on translating hypotheses concerning brain disorders into features to be used in predictive models.