Abstract：
Within-host models describe the dynamics of immune cells when encountering a pathogen, and how these dynamics can lead to an individual-specific immune response. This systematic review aims to summarize which within-host methodology has been used to study and quantify antibody kinetics after infection or vaccination. In particular, we focus on data-driven and theory-driven mechanistic models.
PubMed and Web of Science databases were used to identify eligible papers published until May 2022. Eligible publications included those studying mathematical models that measure antibody kinetics as the primary outcome (ranging from phenomenological to mechanistic models).
We identified 78 eligible publications, of which 8 relied on an Ordinary Differential Equations (ODEs)-based modelling approach to describe antibody kinetics after vaccination, and 12 studies used such models in the context of humoral immunity induced by natural infection. Mechanistic modeling studies were summarized in terms of type of study, sample size, measurements collected, antibody half-life, compartments and parameters included, inferential or analytical method, and model selection.
Despite the importance of investigating antibody kinetics and underlying mechanisms of (waning of) the humoral immunity, few publications explicitly account for this in a mathematical model. In particular, most research focuses on phenomenological rather than mechanistic models. The limited information on the age groups or other risk factors that might impact antibody kinetics, as well as a lack of experimental or observational data remain important concerns regarding the interpretation of mathematical modeling results. We reviewed the similarities between the kinetics following vaccination and infection, emphasising that it may be worth translating some features from one setting to another. However, we also stress that some biological mechanisms need to be distinguished. We found that data-driven mechanistic models tend to be more simplistic, and theory-driven approaches lack representative data to validate model results.
PubMed and Web of Science databases were used to identify eligible papers published until May 2022. Eligible publications included those studying mathematical models that measure antibody kinetics as the primary outcome (ranging from phenomenological to mechanistic models).
We identified 78 eligible publications, of which 8 relied on an Ordinary Differential Equations (ODEs)-based modelling approach to describe antibody kinetics after vaccination, and 12 studies used such models in the context of humoral immunity induced by natural infection. Mechanistic modeling studies were summarized in terms of type of study, sample size, measurements collected, antibody half-life, compartments and parameters included, inferential or analytical method, and model selection.
Despite the importance of investigating antibody kinetics and underlying mechanisms of (waning of) the humoral immunity, few publications explicitly account for this in a mathematical model. In particular, most research focuses on phenomenological rather than mechanistic models. The limited information on the age groups or other risk factors that might impact antibody kinetics, as well as a lack of experimental or observational data remain important concerns regarding the interpretation of mathematical modeling results. We reviewed the similarities between the kinetics following vaccination and infection, emphasising that it may be worth translating some features from one setting to another. However, we also stress that some biological mechanisms need to be distinguished. We found that data-driven mechanistic models tend to be more simplistic, and theory-driven approaches lack representative data to validate model results.
摘要：
背景：宿主内模型描述了遇到病原体时免疫细胞的动力学，以及这些动力学如何导致个体特异性免疫反应。本系统综述旨在总结哪些宿主内方法已用于研究和定量感染或疫苗接种后的抗体动力学。特别是,我们专注于数据驱动和理论驱动的机制模型。
方法：使用PubMed和WebofScience数据库来识别直到2022年5月发表的合格论文。合格的出版物包括研究测量抗体动力学作为主要结果的数学模型（从现象学到机械模型）。
结果：我们确定了78个合格出版物，其中8个依赖于基于常微分方程（ODE）的建模方法来描述疫苗接种后的抗体动力学，和12项研究在自然感染诱导的体液免疫的背景下使用了这样的模型。从研究类型的角度总结了机械建模研究，样本量,收集的测量，抗体半衰期，包括隔室和参数，推理或分析方法，和模型选择。
结论：尽管研究体液免疫（减弱）的抗体动力学和潜在机制很重要，很少有出版物在数学模型中明确说明了这一点。特别是,大多数研究集中在现象学而不是机械学模型。关于年龄组或其他可能影响抗体动力学的危险因素的信息有限，以及缺乏实验或观测数据仍然是关于数学建模结果解释的重要问题。我们回顾了疫苗接种后的动力学和感染之间的相似性，强调将一些功能从一个设置转换到另一个设置可能是值得的。然而,我们还强调需要区分一些生物学机制。我们发现数据驱动的机械模型往往更简单，理论驱动的方法缺乏验证模型结果的代表性数据。
方法：使用PubMed和WebofScience数据库来识别直到2022年5月发表的合格论文。合格的出版物包括研究测量抗体动力学作为主要结果的数学模型（从现象学到机械模型）。
结果：我们确定了78个合格出版物，其中8个依赖于基于常微分方程（ODE）的建模方法来描述疫苗接种后的抗体动力学，和12项研究在自然感染诱导的体液免疫的背景下使用了这样的模型。从研究类型的角度总结了机械建模研究，样本量,收集的测量，抗体半衰期，包括隔室和参数，推理或分析方法，和模型选择。
结论：尽管研究体液免疫（减弱）的抗体动力学和潜在机制很重要，很少有出版物在数学模型中明确说明了这一点。特别是,大多数研究集中在现象学而不是机械学模型。关于年龄组或其他可能影响抗体动力学的危险因素的信息有限，以及缺乏实验或观测数据仍然是关于数学建模结果解释的重要问题。我们回顾了疫苗接种后的动力学和感染之间的相似性，强调将一些功能从一个设置转换到另一个设置可能是值得的。然而,我们还强调需要区分一些生物学机制。我们发现数据驱动的机械模型往往更简单，理论驱动的方法缺乏验证模型结果的代表性数据。
