Biography

Biostatistician Noorie Hyun, PhD, collaborates on projects across a range of research at Kaiser Permanente Washington Health Research Institute, including addiction disorder, pragmatic clinical trials, cervical cancer screening, and drug and vaccine safety and effectiveness. Her current research focuses on designing and incorporating validation data to improve the precision of error-prone data from electronic health records and surveys and developing risk prediction models for personalized disease screening triage.

Before joining KPWHRI, Dr. Hyun worked as an assistant professor in the Division of Biostatistics and the Center for Advancing Population Science at the Medical College of Wisconsin. She provided scientific leadership, from study designs for observational studies and clinical trials through statistical analysis. Dr. Hyun also collaborated with the Surveillance and Health Equity researchers in the American Cancer Society to evaluate risks for cancer survivors compared to the general population as health disparity research.

Dr. Hyun received post-doctoral fellowship training in the Division of Cancer Epidemiology and Genetics (DCEG) at the National Cancer Institute. She developed statistical risk prediction models to estimate cervical cancer prevalence and incidence while addressing complex data-driven issues in large electronic health record systems such as the Kaiser Permanente Northern California cervical cancer screening cohort. Dr. Hyun also developed effective group testing methods for estimating prevalences of categorical traits and estimating the kappa statistic for agreement between 2 ordinal outcomes for multi-stage cluster sample data. All the developed methods were motivated by data-driven problems arising from collaborations with the DCEG epidemiologists.

Dr. Hyun received her PhD degree in biostatistics from the University of North Carolina at Chapel Hill. Her dissertation work focused on flexible semiparametric models to account for the measurement error in longitudinal biomarker outcome measurements and skewed distribution with a thick tail for the biomarker outcome. In the models, properly fitting the tail part of the distribution corresponding to abnormal/extreme blood glucose values as a diabetes biomarker detects well the association between biomarker and patient risk factors.