Grant: Project Grant

Health outcomes such as disease occurrence or stroke are related to many behavioral, clinical and environmental factors, and one of the main priorities of population health research is to explore these complex relationships. The advent of efficient methods for clinical data collection, and of large longitudinal studies that follow participants over several years, create exciting new opportunities to get more in-depth insights into the dynamic processes affecting human health. From this perspective, especially valuable data include repeated measurements of the same potential risk factor, such as blood pressure, at different times. However, on the other hand, such rich longitudinal data impose also important challenges for data analysts that attempt to unravel the underlying biological and clinical relationships through rigorous statistical analyses. To address these complexities, we propose to develop new, flexible statistical models to assess how a given health outcome depends on past and more recent values of longitudinal risk factors. The proposed new methods will be first validated in extensive computer simulations, which - in statistical research - play a role similar to scientific experiments in bench-based fundamental medical or pharmaceutical research. Next, working together with collaborators who have expertise in different fields of clinical or population health research, we will employ these new statistical models in several real-world longitudinal studies of different diseases. Recently, such interdisciplinary collaborations have demonstrated how our flexible methods allowed a more accurate assessment of the therapeutic effectiveness or secondary adverse effects of various medications. Accordingly, we expect that the currently proposed new methods will help understand how cumulative effects of values of risk factors such as air pollution or elevated blood pressure affect the risk of developing a cancer or cardiovascular disease.