Activity Number:
|
69
- Longitudinal/Correlated Data II
|
Type:
|
Contributed
|
Date/Time:
|
Sunday, July 29, 2018 : 4:00 PM to 5:50 PM
|
Sponsor:
|
Biometrics Section
|
Abstract #330799
|
Presentation
|
Title:
|
Bayesian Multivariate Longitudinal Models for Bariatric Surgery Outcomes
|
Author(s):
|
Heidi Fischer* and Karen Coleman and Robert Weiss and Stephen Derose and Allon Friedman and David H. Smith and Talha Imam
|
Companies:
|
Kaiser Permanente Southern California and Kaiser Permanente Southern California and UCLA and Kaiser Permanente Southern California and Indiana University School of Medicine and Kaiser Permanente Center for Health Research and Kaiser Permanente Southern California
|
Keywords:
|
longitudinal analysis;
multivariate mixed models;
Bayesian statistics;
bariatric surgery
|
Abstract:
|
Though bariatric surgery has been shown to result in both weight loss and improved kidney function, little is known about how bariatric surgery may alter the relationship between weight and glomerular filtration rate (eGFR) and if that relationship changes over time. We propose modeling eGFR and weight trajectories jointly before and after bariatric surgery using Bayesian multivariate functional generalized linear mixed models. Weight and eGFR trajectories are modeled over time with a cubic B-spline basis. Association between outcomes is generated by allowing random components of these trajectories to be correlated and correlation coefficients are estimated over time. Model results are communicated graphically. These methods are highly flexible, supporting multiple outcomes with measurements occurring at different time points, missing data, and the flexible modeling of a wide array of mean and association structures. The Bayesian paradigm allows the estimation of a wide range of statistical quantities together with 95 percent credible intervals.
|
Authors who are presenting talks have a * after their name.