UC San Diego

We are at a pivotal point in healthcare informatics where our technical abilities and computational methods are often beyond our abilities to translate and implement them in practical and applied settings. For example, researchers are working on unraveling the intricacies of the human genome to help individuals with their healthcare needs, but often lack the ability to easily share genetic data and medical records between clinics or healthcare providers due to logistical, legal, and interoperability issues. Using more sophisticated healthcare data on individuals derived from other data-intensive assays, such imaging or digital devices, to guide their care is likely to create even more practical issues. This will especially be the case as the community pushes towards personalized medicine and an ability to track an individual’s health over time through the use of these data-intensive assays. In my thesis research, I consider the application of statistical methods to large-scale longitudinal health data. I focus on not only the detection of novel relationships between longitudinal data points, but also on developing strategies for making more personalized health predictions and recommendations. I focus on three broad settings: one involving an analysis of genetic factors associated with progressive visual field loss in patients with primary open angle glaucoma using a large clinical data set, and two that consider individual mental health and mood data obtained from the prolonged use of a digital therapeutic (i.e., smartphone app) designed to provide meditations to individual users to reduce stress, anxiety, and depression. I also discuss and explore the implementation of systems designed to learn or adapt from past data in order to improve predictive models going forward using the glaucoma and healthcare app data.