Patients suffering from Mycobacterium tuberculosis (MTB) infections frequently require lengthy (18-24 months) treatments with multiple drugs. Adverse drug events in addition to high death rates and therapeutic failures represent a substantial challenge for these patients. A major hypothesis of this project is that drug regimen sequencing will markedly shorten the length of TB treatments and prevent the emergence of resistance. To test this hypothesis, we will use three different preclinical experimental systems as well as metabolic state data of MTB: i) to identify switch regimens that target MTB populations that have been metabolically altered and made less responsive to drug treatment over time due to the primary regimen and ii) to select follow- on regimens that have resistance mechanisms that are independent of those employed by the primary regimen. The mathematical modeling core will play an essential role for the successful achievement of these goals as it will allow for the integration of data from all three projects and the Assay Core through the use of cutting edge mathematical modeling and simulation approaches into an overarching mathematical framework. Once established and qualified, this mathematical framework will allow us to compare and contrast different drug regimens in terms of bacterial cell kill and suppression of resistance and, hence, serve as the engine that drives the whole proposal. Data used for this innovative analysis will originate from in vitro (Project #1) and in vivo animal systems (Project #2: Mouse; Project #3: cynomolgus macaque), which will be used to identify the most promising drug and dosing regimen that will (hopefully) effectively treat MTB infections in humans in a more expeditious and rational manner as follows: 1) concentrations representative of the surrogate target site, i.e. ELF concentrations, will be characterized for all three projects using a population pharmacokinetic (pop-PK) analysis approach and 2) linked to their corresponding antimicrobial effect considering different susceptibilities to drugs for different bacterial subpopulations. Finally, simple and Monte Carlo simulations will be performed to determine the effects of between-subject variability on cell kill and resistance suppression. The non-parametric adaptive grid (NPAG) algorithm within Pmetrics will provide the computational engine behind this proposal. As part of the mathematical core activities, NPAG will be enhanced with new parallel computational capabilities, as well as the ability to model data that are categorical, discrete, or time-to-event. These updates will increase the speed of the analysis and extend it to properly handle bacterial colony counts and animal survival data generated in Projects 1-3.