WITHDRAWN: Bayesian Modeling and Prediction of Patient Accrual Using Gaussian Processes

• Activity Number: 66
• Type: Topic Contributed
• Date/Time: Sunday, August 3, 2014 : 4:00 PM to 5:50 PM
• Sponsor: Social Statistics Section
• Author(s): Qing He and Chandra Thames and Jeen Liu and Qi Long
• Companies: Emory University and Eisai and Eisai and Emory University
• Keywords: Clinical Trial ; Nonparametric Bayesian ; Accrual Prediction ; Gaussian Process ; Multi-region ; non-homogeneous Poisson process

Abstract

Ensuring adequate patient recruitment including allocation of sufficient time for recruitment is a crucial aspect of planning and monitoring a clinical trial. In particular, prediction of both the expected recruitment by a future time point and the accrual period needed to reach a target recruitment goal is often of interest. Existing methods either model patient accrual as a Poisson process with a fixed accrual rate (Senn, 1998)or as piecewise linear or constant with pre-specified or estimated change points (Tang et al, 2012), or use non-homogeneous Poisson process to model patient accrual, where a smooth spline function is adopted for the underlying accrual rate (Zhang and Long, 2010). Our method assumes a time varying accrual rate which is modeled using Gaussian process (GP). GP applies to a great variety of accrual patterns and is suitable to model accrual rates of multi-regional trials and with different initiation times. Our numerical studies show that the proposed method improves accuracy and precision of accrual prediction compared to existing methods.