Predicting infectious disease dynamics is a central challenge in disease ecology. Models that can assess which individuals are most at risk of being exposed to a pathogen not only provide valuable insights into disease transmission and dynamics but can also guide management interventions. Constructing such models for wild animal populations, however, is particularly challenging; often only serological data are available on a subset of individuals and nonlinear relationships between variables are common. Here we provide a guide to the latest advances in statistical machine learning to construct pathogen‐risk models that automatically incorporate complex nonlinear relationships with minimal statistical assumptions from ecological data with missing data. Our approach compares multiple machine learning algorithms in a unified environment to find the model with the best predictive performance and uses game theory to better interpret results. We apply this framework on two major pathogens that infect African lions: canine distemper virus (CDV) and feline parvovirus. Our modelling approach provided enhanced predictive performance compared to more traditional approaches, as well as new insights into disease risks in a wild population. We were able to efficiently capture and visualize strong nonlinear patterns, as well as model complex interactions between variables in shaping exposure risk from CDV and feline parvovirus. For example, we found that lions were more likely to be exposed to CDV at a young age but only in low rainfall years. When combined with our data calibration approach, our framework helped us to answer questions about risk of pathogen exposure that are difficult to address with previous methods. Our framework not only has the potential to aid in predicting disease risk in animal populations, but also can be used to build robust predictive models suitable for other ecological applications such as modelling species distribution or diversity patterns.
- NSF-PAR ID:
- 10230196
- Editor(s):
- Griffith, Gary
- Date Published:
- Journal Name:
- ICES Journal of Marine Science
- Volume:
- 77
- Issue:
- 4
- ISSN:
- 1095-9289
- Page Range / eLocation ID:
- 1463 to 1479
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
Abstract -
Experiments and models suggest that climate affects mosquito-borne disease transmission. However, disease transmission involves complex nonlinear interactions between climate and population dynamics, which makes detecting climate drivers at the population level challenging. By analysing incidence data, estimated susceptible population size, and climate data with methods based on nonlinear time series analysis (collectively referred to as empirical dynamic modelling), we identified drivers and their interactive effects on dengue dynamics in San Juan, Puerto Rico. Climatic forcing arose only when susceptible availability was high: temperature and rainfall had net positive and negative effects respectively. By capturing mechanistic, nonlinear and context-dependent effects of population susceptibility, temperature and rainfall on dengue transmission empirically, our model improves forecast skill over recent, state-of-the-art models for dengue incidence. Together, these results provide empirical evidence that the interdependence of host population susceptibility and climate drives dengue dynamics in a nonlinear and complex, yet predictable way.more » « less
-
Abstract Experiments and models suggest that climate affects mosquito‐borne disease transmission. However, disease transmission involves complex nonlinear interactions between climate and population dynamics, which makes detecting climate drivers at the population level challenging. By analysing incidence data, estimated susceptible population size, and climate data with methods based on nonlinear time series analysis (collectively referred to as empirical dynamic modelling), we identified drivers and their interactive effects on dengue dynamics in San Juan, Puerto Rico. Climatic forcing arose only when susceptible availability was high: temperature and rainfall had net positive and negative effects respectively. By capturing mechanistic, nonlinear and context‐dependent effects of population susceptibility, temperature and rainfall on dengue transmission empirically, our model improves forecast skill over recent, state‐of‐the‐art models for dengue incidence. Together, these results provide empirical evidence that the interdependence of host population susceptibility and climate drives dengue dynamics in a nonlinear and complex, yet predictable way.
-
Abstract Aim One of the primary characteristics that determines the structure and function of marine food webs is the utilization and prominence of energy‐rich lipids. The biogeographical pattern of lipids throughout the ocean delineates the marine “lipidscape,” which supports lipid‐rich fish, mammal, and seabird communities. While the importance of lipids is well appreciated, there are no synoptic measurements or biogeographical estimates of the marine lipidscape. Productive lipid‐rich food webs in the pelagic ocean depend on the critical diapause stage of large pelagic copepods, which integrate lipid production from phytoplankton, concentrating it in space and time, and making it available to upper trophic levels as particularly energy‐rich wax esters. As an important first step towards mapping the marine lipidscape, we compared four different modelling approaches of copepodid diapause, each representing different underlying hypotheses, and evaluated them against global datasets.
Location Global Ocean.
Taxon Copepoda.
Methods Through a series of global model runs and data comparisons, we demonstrated the potential for regional studies to be extended to estimate global biogeographical patterns of diapause. We compared four modelling approaches each designed from a different perspective: life history, physiology, trait‐based community ecology, and empirical relationships. We compared the resulting biogeographical patterns and evaluated the model results against global measurements of copepodid diapause.
Results Models were able to resolve more than just the latitudinal pattern of diapause (i.e. increased diapause prevalence near the poles), but to also pick up a diversity of regions where diapause occurs, such as coastal upwelling zones and seasonal seas. The life history model provided the best match to global observations. The predicted global biogeographical patterns, combined with carbon flux estimates, suggested a lower bound of 0.031–0.25 Pg C yr−1of downward flux associated with copepodid diapause.
Main conclusions Results indicated a promising path forward for representing a detailed biogeography of the marine lipidscape and its associated carbon flux in global ecosystem and climate models. While complex models may offer advantages in terms of reproducing details of community structure, simpler theoretically based models appeared to best reproduce broad‐scale biogeographical patterns and showed the best correlation with observed biogeographical patterns.
-
Abstract Host–parasite dynamics are impacted by the relationship between host density and parasite transmission, and thus, all epidemiological models contain a central transmission–density function. Recent theoretical work demonstrates that this central parasite transmission function might be best represented by a nonlinear continuum from one linear extreme to another: density‐dependent transmission at low host densities to density‐independent transmission at high host densities. But how often are nonlinear transmission functions used, and when are they better at describing transmission in real host–parasite systems?
To quantify existing modelling practices, we systematically reviewed seven representative ecology journals, finding 262 studies containing host–parasite models that contained linear and/or nonlinear transmission functions. We also reviewed the literature to find 28 experimental and observational studies that compared multiple transmission functions in real host–parasite systems, and tallied which functions were best supported in those systems. Finally, we created a flexible model simulation tool to explore and quantify the bias in model parameter estimates that is created when using an inaccurate transmission function.
We found that most experimental and observational studies reported that nonlinear transmission–density functions outperformed simple linear transmission–density functions, supporting recent theoretical work. In contrast, most studies containing host–parasite models assumed that host density was constant and/or used a single, linear transmission function to explain how transmission rates changed with density. Using the wrong linear function and/or using a linear function when the underlying transmission–density relationship is even slightly nonlinear can substantially bias model parameter estimates, as demonstrated by our simulations over a broad parameter space.
Some modelling studies may be using linear functions in host–parasite systems where nonlinear functions are more appropriate. If true, these models would yield substantially biased parameter estimates. To avoid such biases that compromise ecological understanding and prediction, we recommend that future studies compare multiple transmission functions, including nonlinear options, whenever possible.