Recent approaches in computational modelling for controlling pathogen threats

Abstract

In this review, we assess the status of computational modelling of pathogens. We focus on three disparate but interlinked research areas that produce models with very different spatial and temporal scope. First, we examine antimicrobial resistance (AMR). Many mechanisms of AMR are not well understood. As a result, it is hard to measure the current incidence of AMR, predict the future incidence, and design strategies to preserve existing antibiotic effectiveness. Next, we look at how to choose the finite number of bacterial strains that can be included in a vaccine. To do this, we need to understand what happens to vaccine and non-vaccine strains after vaccination programmes. Finally, we look at within-host modelling of antibody dynamics. The SARS-CoV-2 pandemic produced huge amounts of antibody data, prompting improvements in this area of modelling. We finish by discussing the challenges that persist in understanding these complex biological systems.

Figure 1.. A model of N. gonorrhoeae infection in men who have sex with men can be used to compare different strategies for the introduction of a new antibiotic.

(A) Schematic of the transmission model, adapted from Reichert et al (2023). Men can be either susceptible (S), asymptomatically infected (Y), or symptomatically infected (Z). Infected men can carry a strain that is resistant to either none, one, or both antibiotics A and B (subscripts). Each compartment contains three subcompartments for men with different levels of sexual activity. The flow between compartments is illustrated by arrows. (B) Illustrative comparison of predictions arising from the model for two possible treatment strategies: keeping B in reserve until resistance levels to A reach a 5% prevalence threshold then switching to B or using A and B in combination from the start. The model predicts a substantial delay in the onset of resistance with the combination strategy, resulting in increased lifespan for both antibiotics. Data taken from Table 2 of original paper. For full details, see the original paper.

Figure 2.. An illustration of vaccine-induced population dynamics.

In each case, an effective vaccine is rolled out against the dominant strain, in yellow. After vaccination, cases of the dominant strain decrease. However other strains, having less competition for hosts, are able to fill the “gap” which has been left in the population. (A) In the first example (panel (A)), a benign non-disease strain (green) fills this gap and therefore total cases of disease in the population are reduced (illustrative of H. influenzae–vaccination against serotype B). (B) In the second example (panel (B)), a more virulent strain is able to take over, ultimately meaning the vaccine has no effect, or even a detrimental effect on, total disease burden (illustrative of N. meningitidis–yellow strain serotype C; red strain serotype W). Modelling can be used to try and predict which situation will occur. (C) In the third example (panel (C)), representing a complex population with many strains (illustrative of Streptococcal pathogens), vaccination against all strains is not feasible. Modelling can be used to predict which non-vaccine variants will come to dominate, whether newly emerged strains will be successful, and optimise the initial vaccine formulation.

Figure 3.. Schematic of a typical antibody kinetics model, fit to longitudinal neutralising antibody data for SARS-CoV-2 stratified by two variants in a hierarchical framework.

(A) A single individuals’ neutralising antibody data (left) and resulting model fits (right), stratified by variant neutralised, using a typical six-parameter model of waning immunity. (B) Hierarchical structure of the model. Top row: examples for three individual trajectories, split by variance. Bottom row: individual-level and population-level kinetics pooled for three individuals, giving more accurate population estimates.