Bayesian Modeling for Change-Points Detection in Longitudinal Clinical Proteomics Experiments

• Activity Number: 450
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 3:05 PM to 3:50 PM
• Sponsor: Quality and Productivity Section
• Author(s): Xia Wang*
• Companies: University of Cincinnati
• Keywords: Change Point ; Bayesian method ; Hidden Markov model ; Proteomics

Abstract

Proteomics technology provides detailed inventories of proteins from a biological sample. It exhibits great potential to help researchers differentiate biological differences between phenotypes (for example, cancer patients vs. healthy individuals). The spread of this technology, however, is accompanied by persistent concerns about the reproducibility of proteomics methods. Statistical quality control is important to maintain the instrument stability. One aim of statistical quality control is to detect the assignable causes that lead to out-of-control process, such as the shift of means or variances and both parameter values. A Bayesian Hidden Markov Model is proposed in detecting the change point in simulation studies and a clinical proteomics experiment data. The proposed methods are compared to binary segmentation, segment neighborhood, and the PELT algorithm.