MATHEMATICAL MODELING OF ANTIMICROBIAL RESISTANCE: AN OVERVIEW

Antimicrobial resistance (AMR) is a critical public health problem that compromises the health and well-being of humans and animals. Mathematical modeling is essential to predict trends, clarify causal relationships, and evaluate intervention scenarios. Objective: we sought to provide a comprehensive overview of mathematical modeling of AMR as a whole, characterizing the types of models (deterministic, stochastic, agent-based, etc) and their modeling goals (pathogen evolution, mortality/morbidity in human populations, etc). Methods: we conducted a non-exhaustive literature search, keeping track of model type and goals. Results: we analyzed fourteen articles, with varied contexts: four in hospitals, four explored within-host microbial dynamics, one investigated human-animal-environment interactions, one in laboratory environments, and four articles had a general context. Immune factors were mentioned in three (21%) of the studies. Only one model simultaneously included pharmacokinetic (PK) and pharmacodynamic (PD) mechanisms. In twelve studies, a compartmental model was used and most models were deterministic against only two stochastic models. No study investigated agent-based models (ABM). Parameter sensitivity analysis was carried out in only six studies. Observed data were reproduced in five studies. Only 21% of the models incorporated the fitness cost associated with AMR. The acquisition of resistance by pathogens, when mentioned, was by horizontal gene transfer (HGT) via conjugation (eight studies) or by mutations (two studies). Conclusions: our findings suggest several important gaps in the literature, mainly the lack of stochastic models that simultaneously account for fitness cost, intra-host dynamics and realistic resistance-acquisition mechanisms.