Toxicokinetic-based survival model for analysis of toxicity data with changing dose

Abstract

<p>In toxicology studies, some data sets are dose-time-response data with changing doses over time and death/alive as response. We developed a toxicokinetic-based survival model that relates survivorship and bioaccumulation. The model assumes toxicant uptake and elimination follow a single compartment model and the hazard rate is proportional to the internal concentration. Several diagnostics methods were described for this model to check assumptions of proportional hazard and the functional form of covariates. Maximum likelihood and least square estimation were compared by comparing the estimating equations and a simulation study. The results showed that maximum likelihood is the most unbiased and efficient estimation method for this model.</p>