Bipolar disorder is a leading cause of disability and death in women of childbearing age and episodes of bipolar disorder immediately following childbirth are common and clinically important. With its discrete, identifiable trigger, bipolar disorder triggered by childbirth represents a unique opportunity for risk prediction modeling in psychiatry.Genetic variants have been robustly implicated in the susceptibility to bipolar disorder. Comprehensive approaches that integrate information on genetic susceptibility with clinical risk factors have been successfully applied to prediction modeling of many disorders and are needed in psychiatry. In this fellowship I plan to bring these techniques to psychiatry and to apply them to the prediction of episodes of bipolar disorder following childbirth.My primary goal is to develop clinically useful predictive models for episodes of bipolar disorder triggered by childbirth. I anticipate that the predictive models will help in individualizing risk assessment, and thus improve the identification of women at risk and the clinical outcomes. In addition, by helping to inform understanding of the interplay of risk factors in pathogenesis the work will contribute towards more focused and effective clinical management.This fellowship will build in two important ways on my PhD research in which I studied postpartum episodes in the large Cardiff bipolar disorder clinical datasets available at the time (N=2,032). First, I will develop and externally validate multivariable prediction models for postpartum bipolar disorder using very large international collaborative samples of over 20,000 individuals.Second, I will receive training in new areas of research that will include multidisciplinary approaches to postpartum mood disorders (in Chapel Hill, North Carolina, USA), biostatistics and bioinformatics (in Cardiff, Wales and Chapel Hill).