Towards clinical relevance: Predictive modeling applied to brain imaging
Abstract
The application of machine learning methods to the increasingly larger observational studies of neurological and psychiatric disorders provides the opportunity of generating individualized assessments for people with a given condition, while also enabling the detection of brain correlates of disease. This program comes with its own challenges: ensuring predictions are robust across studies, dealing with the heterogeneity of disease expression, and providing interpretable results that are useful in a clinical context. In this presentation, I will show a few applications of machine learning models to brain imaging data that take into account these issues. In particular, I will focus on disease trajectory characterization and population stratification.
Speaker
Eduardo is part of the Computational Psychiatry and Neuroimaging group at the Healthcare and Life Sciences department. He applies his background in machine learning to apply models to brain imaging in order to better understand the dynamics of neurological and psychiatric disorders. In doing so, a number of open issues in the field could be addressed, such as identifying heterogeneity in clinical populations or assessing risk for severe disease. In summary, his goal is to use predictive models to exploit the full clinical potential of brain imaging.
Requested 1 time