Serodynamics: A primer and synthetic review of methods for epidemiological inference using serological data

Abstract

We present a review and primer of methods to understand epidemiological dynamics and identify past exposures from serological data, referred to as serodynamics. We discuss processing and interpreting serological data prior to fitting serodynamical models, and review approaches for estimating epidemiological trends and past exposures, ranging from serocatalytic models applied to binary serostatus data, to more complex models incorporating quantitative antibody measurements and immunological understanding. Although these methods are seemingly disparate, we demonstrate how they are derived within a common mathematical framework. Finally, we discuss key areas for methodological development to improve scientific discovery and public health insights in seroepidemiology.

Keywords: Infectious disease modeling; Serodynamics; Seroepidemiology; Serology.

Fig. 1

From blood vials to processed seroepidemiological data and sources of variation in the observation process. Tables include upstream considerations in the measurement process (i.e., testing procedures, normalization and standardization of lab data, and classification of serostatus) that can affect downstream serodynamics applications.

Fig. 2

Overview of serodynamics methods for reconstructing transmission dynamics. Different serodynamics methods can be used to estimate transmission dynamics depending on the type and availability of data. For each approach (row), the figure shows data requirements, a depiction of the model structure, and key model outputs. (A) Serocatalytic models require only binary serostatus and age to link age-stratified seroprevalence to the FOI, optionally accounting for additional complexity such as seroreversion. (B) Antibody acquisition models have a similar structure to serocatalytic models, but using antibody level rather than serostatus. (C) Simple TSI methods use quantitative antibody levels combined with a model of expected antibody kinetics to back-calculate an individual’s likely infection time from one or more measurements. This simple approach can be extended to account for co-circulation, reinfection and cross-reactivity of multiple pathogens and measurement of multiple biomarkers, enabling fitting of more complex models (D). (E) Machine learning approaches rely on similar input data and may be used to answer similar questions regarding TSI, but do not explicitly model the data-generating process.