NIH funds development of novel biostatistics method for multilevel time-dynamic modeling
February 19, 2016
A joint UC Irvine and UCLA research proposal to develop novel biostatistics methodology, led by Drs. Danh Nguyen (Director, UCI ICTS Biostatistics, Epidemiology and Research Design Unit) and Damla Senturk (Professor in the Department of Biostatistics, UCLA Fielding School of Public Health), was funded by a four-year NIH/NIDDK grant. The research program will address gaps in statistical methodology for assessing time-varying effects. For example, characterizing the time-dynamic (time-varying) effects of risk factors on outcomes of patients with chronicdiseases, such as cardiovascular (CV) events and infection in patients on dialysis, is important for exploring more effective approaches to disease management and prevention. In particular, potentially more effective CV risk reduction strategies will require understanding the complex time-dynamic changes in patients’ CV risk trajectories over time to allow for identification of timeframes of increased CV risk. The research program will advance a general framework to estimation and inference for multilevel time-dynamic modeling that accommodates multilevel data structures (e.g., patients nested within dialysis facilities, hospitals or care providers, and observations over time nested within patients). Our proposed multilevel varying coefficient modeling framework will address the specific dual sets of goals: 1) patient-level inference, such as patient-centered decision making and understanding the time-varying effects of patient risk factors on outcomes over time; and 2) time-dynamic (provider-) facility-level inference, including quantification of facility-level factors’ effects on patient outcome, prediction and assessment of facility performance with appropriate patient risk, case-mix adjustment. Other co-investigators include Drs. Kam Kalantar-Zadeh and Connie Rhee in the UCI Division of Nephrology.