Estimation and Prediction of Longitudinal Biomarker Distributions Using Bayesian Nonparametric Beta Regression

• Activity Number: 616
• Type: Contributed
• Date/Time: Wednesday, August 12, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Shouhao Zhou* and Xuelin Huang
• Companies: MD Anderson Cancer Center and MD Anderson Cancer Center
• Keywords: Longitudinal Analysis ; Bayesian Beta Regression ; Fractional Polynomials ; Chronic Myeloid Leukemia ; BCR-ABL

Abstract

In this paper we present an innovative non-parametric Bayesian approach to estimate and predict individual patient longitudinal biomarker distributions. It was motivated by a study for chronic myeloid leukemia, in which BCR-ABL expression level is a very good indicator of disease residual after the treatment of tyrosine kinase inhibitors. We propose a non-parametric method using Bayesian hierarchical modelling to estimate smooth curves of BCR-ABL level distribution over time jointly with other clinical factors, such that physicians can not only cross-sectionally understand how extreme a recently measured BCR-ABL level for CML patients, but also predict the future change of BCR-ABL levels for any specific CML patient given the historical BCR-ABL levels.