Clustering Functional Databased on Change-Point Models for Electronic Monitoring Data

• Activity Number: 139 - Challenges and Advances in Statistical Inference for Problems with Nonregularity in the Era of Big Data
• Type: Invited
• Date/Time: Monday, July 31, 2017 : 10:30 AM to 12:20 PM
• Sponsor: WNAR
• Author(s): Chongzhi Di* and Yifan Zhu and Ying Qing Chen
• Companies: Fred Hutchinson Cancer Research Center and Fred Hutchinson Cancer Research Center and Fred Hutchinson Cancer Research Center
• Keywords: clustering ; functional data ; change points ; adherence

Abstract

We consider the problem of clustering functional data arising from electronic monitoring devices to monitor adherence in clinical trials. Our motivation study is HPTN 069, a phase II randomized, double-blind PrEP study to investigate adherence patterns in HIV prevention trials. The study utilized Wisepill, an electronic drug monitoring device to collect daily pill bottle opening signals in a one-year study period to study adherence. We propose both parametric models based on change points and nonparametric approaches based on functional data analysis for clustering adherence patterns. We also develop bias correction methods for subsequent regression models, which relates participants' characteristics to their types of adherence patterns. These methods are applied to the HIV 069 and offer new insights of adherence patterns.