Neurological and psychiatric disorders pose a large burden on the individual, their families and our society at large. An increasingly aging population makes it imperative that we detect these crippling disorders at an early stage, giving the subject a greater chance at recovery and reduce the cost of healthcare for our society. Brain disorders are linked to abnormal patterns of brain function and structure.
We will use different machine learning techniques to develop models that can predict disease development on the basis of structural and functional imaging and genetic data, which can serve as biomarkers for brain disorders. Identifying individuals at risk of developing severe brain disorders gives us the best chance to intervene at an early stage of the disorder. This could dramatically increase the chance of the individuals having a better quality-of-life and reduce the negative impact on their, family, friends, health care workers and society. We will analyse the structural brain images from the individuals from the UKBiobank to build a deep learning network. This network will be trained to 1) predict the genetic liabilities for specific psychiatric/brain disorders that will be calculated using the genetic information from these individuals, and 2) the development of diseases during follow-up. We will include all individuals with structural (T1, T2, DTI) and/or functional (resting-state, task-based) datasets along with their genetic information whenever available.
We will also include (non-imaging) data from all other participants from whom genetics is available, in order to create the covariates for further analysis based on better estimates of the population structure.