Inferring random change point from left-censored longitudinal data by segmented mechanistic nonlinear models, with application in HIV surveillance study
The primary goal of public health efforts to control HIV epidemics is to diagnose and treat people with HIV infection as soon as possible after seroconversion. The timing of initiation of antiretroviral therapy (ART) treatment after HIV diagnosis is, therefore, a critical population-level indicator that can be used to measure the effectiveness of public health programs and policies at local and national levels. However, population-based data on ART initiation are unavailable because ART initiation and prescription are typically measured indirectly by public health departments (e.g., with viral suppression as a proxy). In this paper, we present a random change-point model to infer the time of ART initiation utilizing routinely reported individual-level HIV viral load from an HIV surveillance system. To deal with the left-censoring and the nonlinear trajectory of viral load data, we formulate a flexible segmented nonlinear mixed effects model and propose a Stochastic version of EM (StEM) algorithm, coupled with a Gibbs sampler for the inference. We apply the method to a random subset of HIV surveillance data to infer the timing of ART initiation since diagnosis and to gain additional insights into the viral load dynamics. Simulation studies are also performed to evaluate the properties of the proposed method.