Emulator-based Bayesian optimization for efficient multi-objective calibration of an individual-based model of malaria

Individual-based models have become important tools in the global battle against infectious diseases, yet model complexity can make calibration to biological and epidemiological data challenging. We propose using a Bayesian optimization framework employing Gaussian process or machine learning emulator functions to calibrate a complex malaria transmission simulator. We demonstrate our approach by optimizing over a high-dimensional parameter space with respect to a portfolio of multiple fitting objectives built from datasets capturing the natural history of malaria transmission and disease progression. Our approach quickly outperforms previous calibrations, yielding an improved final goodness of fit. Per-objective parameter importance and sensitivity diagnostics provided by our approach offer epidemiological insights and enhance trust in predictions through greater interpretability.

Spatial scale of transmission explains how acute infections circulate at low prevalence

Fundamental questions remain about the regulation of acute pathogens in the absence of acquired immunity. This is especially true for canine rabies, a universally fatal zoonosis. From tracing rabies transmission in a population of 50,000 dogs in Tanzania between 2002-2016 we unravel the processes through which rabies is regulated and persists, fitting individual-based models to spatially-resolved data to investigate the mechanisms modulating transmission and the scale over which they operate. We find that while prevalence never exceeds 0.15%, we detect significant susceptible depletion at local scales commensurate with rabid dog movement, reducing transmission through clustering of rabies deaths and individuals incubating infection. Individual variation in rabid dog behaviour facilitates virus dispersal and co-circulation of lineages, enabling metapopulation persistence. These mechanisms likely operate in many pathogens circulating in spatially structured populations, with important implications for prediction and control, yet are unobservable unless the scale of host interactions is identified.

Individual-based eco-evolutionary models for understanding adaptation in changing seas

As climate change threatens species' persistence, predicting the potential for species to adapt to rapidly changing environments is imperative for the development of effective conservation strategies. Eco-evolutionary individual-based models (IBMs) can be useful tools for achieving this objective. We performed a literature review to identify studies that apply these tools in marine systems. Our survey suggested that this is an emerging area of research fuelled in part by developments in modelling frameworks that allow simulation of increasingly complex ecological, genetic and demographic processes. The studies we identified illustrate the promise of this approach and advance our understanding of the capacity for adaptation to outpace climate change. These studies also identify limitations of current models and opportunities for further development. We discuss three main topics that emerged across studies: (i) effects of genetic architecture and non-genetic responses on adaptive potential; (ii) capacity for gene flow to facilitate rapid adaptation; and (iii) impacts of multiple stressors on persistence. Finally, we demonstrate the approach using simple simulations and provide a framework for users to explore eco-evolutionary IBMs as tools for understanding adaptation in changing seas.

Individual-based models

Individual-based models (IBMs, also known as agent-based models) are mechanistic models in which demographic population trends emerge from processes at the individual level. IBMs are used instead of more aggregated approaches whenever one or more of the following aspects are deemed too relevant to be ignored: intraspecific trait variation, local interactions, adaptive behaviour, and response to spatially and temporally heterogeneous environments, which often results in nonlinear feedbacks. IBMs offer a high degree of flexibility and therefore vary widely in structure and resolution, depending on the research question, system under investigation, and available data. Data used to parameterise an IBM can be divided into two categories: species and environmental data. Unlike other model types, qualitative empirical knowledge can be taken into account via probabilistic rules. IBM flexibility is often associated with higher number of parameters and hence higher uncertainty; therefore sensitivity analysis and validation are extremely important tools for analysing these models. The chapter presents three examples: a vole–mustelid model used to understand the mechanisms underlying population cycles in rodents; a wild boar–virus model to study persistence of wildlife diseases in heterogeneous landscapes; and a wild tobacco-moth caterpillar model to study emergence of delayed chemical plant defence against insect herbivores. These examples demonstrate the ability of IBMs to decipher mechanisms driving observed phenomena at the population level and their role in planning applied conservation measures. IBMs typically require more data and effort than other model types, but rewards in terms of structural realism, understanding, and decision support are high.

Demographic Methods across the Tree of Life

Demography is everywhere in our lives: from birth to death. Demography shapes our daily decisions, as well as the decisions that others make on us (e.g. bank loans, retirement age). Demography is everywhere across the Tree of Life. The universal currencies of demography—survival, development, reproduction, and recruitment—shape the performance of all species, from lions to dandelions. The omnipresence of demography in all things alive and dead, and its multiple applications to better understand the ecology, evolution, and conservation/management of species, allows us to—in principle—apply the wide array of quantitative methods to, for example, bacteria or humans. However, demographic methods to date have remained taxonomically siloed, despite the fact that, to a large extent, they are widely applicable across the Tree of Life. In this book, we walk nonexperts through the ABCs of data collection, model construction, analyses, and interpretation across a wide repertoire of demographic artillery. This book introduces the reader to some of the demographic methods, including abundance-based models, life tables, matrix population models, integral projection models, integrated population models, and individual based models, to mention a few. Through the careful integration of data collection methods, analytical approaches, and applications, clearly guided through fully reproducible R scripts, we provide a state-of-the-art thorough representation of many of the most popular tools that any demographer (or demographically inclined mind) should equip themselves with.

Comparison of two simulators for individual based models in HIV epidemiology in a population with HSV 2 in Yaoundé (Cameroon)

Model comparisons have been widely used to guide intervention strategies to control infectious diseases. Agreement between different models is crucial for providing robust evidence for policy-makers because differences in model properties can influence their predictions. In this study, we compared models implemented by two individual-based model simulators for HIV epidemiology in a heterosexual population with Herpes simplex virus type-2 (HSV-2). For each model simulator, we constructed four models, starting from a simplified basic model and stepwise including more model complexity. For the resulting eight models, the predictions of the impact of behavioural interventions on the HIV epidemic in Yaoundé-Cameroon were compared. The results show that differences in model assumptions and model complexity can influence the size of the predicted impact of the intervention, as well as the predicted qualitative behaviour of the HIV epidemic after the intervention. These differences in predictions of an intervention were also observed for two models that agreed in their predictions of the HIV epidemic in the absence of that intervention. Without additional data, it is impossible to determine which of these two models is the most reliable. These findings highlight the importance of making more data available for the calibration and validation of epidemiological models.