Pharmacogenetic Dosing Algorithm

The expected outputs of the dosing algorithm include ancestry-informed dosing recommendations and a preliminary predictive tool for future clinical validation in Southern African cohorts. External validation of the dosing algorithm will use data from other PoPG sites across Southern Africa. Machine-learning integration in the algorithm will use Bayesian mixed-effects models and stratified linkage disequilibrium score regression to estimate the proportion of dose-response variability explained by genetics, ancestry and clinical covariates. The study plans to develop a pharmacogenetic algorithm for initial TB drug dosing that predicts initiation doses to rapidly achieve therapeutic exposure while reducing time to target concentration. Algorithm development will initially focus on isoniazid and bedaquiline because pharmacogenetic markers are known or suspected to influence their clearance.