An agent-based model of dengue virus transmission shows how uncertainty about breakthrough infections influences vaccination impact projections

Catching a wave: On the suitability of traveling-wave solutions in epidemiological modeling

Ordinary differential equation models such as the classical SIR model are widely used in epidemiology to study and predict infectious disease dynamics. However, these models typically assume that populations are homogeneously mixed, ignoring possible variations in disease prevalence due to spatial heterogeneity. To address this issue, reaction-diffusion models have been proposed as an alternative approach to modeling spatially continuous populations in which individuals move in a diffusive manner. In this study, we explore the conditions under which such spatial structure must be explicitly considered to accurately predict disease spread, and when the assumption of homogeneous mixing remains adequate. In particular, we derive a critical threshold for the diffusion coefficient below which disease transmission dynamics exhibit spatial heterogeneity. We validate our analytical results with individual-based simulations of disease transmission across a two-dimensional continuous landscape. Using this framework, we further explore how key epidemiological parameters such as the probability of disease establishment, its maximum incidence, and its final epidemic size are affected by incorporating spatial structure into SI, SIS, and SIR models. We discuss the implications of our findings for epidemiological modeling and identify design considerations and limitations for spatial simulation models of disease dynamics.

Gaussian Process Emulation for Modeling Dengue Outbreak Dynamics

Epidemiological models that aim for a high degree of biological realism by simulating every individual in a population are unavoidably complex, with many free parameters, which makes systematic explorations of their dynamics computationally challenging. This study investigates the potential of Gaussian Process emulation to overcome this obstacle. To simulate disease dynamics, we developed an individual-based model of dengue transmission that includes factors such as social structure, seasonality, and variation in human movement. We trained three Gaussian Process surrogate models on three outcomes: outbreak probability, maximum incidence, and epidemic duration. These models enable the rapid prediction of outcomes at any point in the eight-dimensional parameter space of the original model. Our analysis revealed that average infectivity and average human mobility are key drivers of these epidemiological metrics, while the seasonal timing of the first infection can influence the course of the epidemic outbreak. We use a dataset comprising more than 1,000 dengue epidemics observed over 12 years in Colombia to calibrate our Gaussian Process model and evaluate its predictive power. The calibrated Gaussian Process model identifies a subset of municipalities with consistently higher average infectivity estimates, highlighting them as promising areas for targeted public health interventions. Overall, this work underscores the potential of Gaussian Process emulation to enable the use of more complex individual-based models in epidemiology, allowing a higher degree of realism and accuracy that should increase our ability to control important diseases such as dengue.

System identifiability in a time-evolving agent-based model

Mathematical models are a valuable tool for studying and predicting the spread of infectious agents. The accuracy of model simulations and predictions invariably depends on the specification of model parameters. Estimation of these parameters is therefore extremely important; however, while some parameters can be derived from observational studies, the values of others are difficult to measure. Instead, models can be coupled with inference algorithms (i.e., data assimilation methods, or statistical filters), which fit model simulations to existing observations and estimate unobserved model state variables and parameters. Ideally, these inference algorithms should find the best fitting solution for a given model and set of observations; however, as those estimated quantities are unobserved, it is typically uncertain whether the correct parameters have been identified. Further, it is unclear what 'correct' really means for abstract parameters defined based on specific model forms. In this work, we explored the problem of non-identifiability in a stochastic system which, when overlooked, can significantly impede model prediction. We used a network, agent-based model to simulate the transmission of Methicillin-resistant staphylococcus aureus (MRSA) within hospital settings and attempted to infer key model parameters using the Ensemble Adjustment Kalman Filter, an efficient Bayesian inference algorithm. We show that even though the inference method converged and that simulations using the estimated parameters produced an agreement with observations, the true parameters are not fully identifiable. While the model-inference system can exclude a substantial area of parameter space that is unlikely to contain the true parameters, the estimated parameter range still included multiple parameter combinations that can fit observations equally well. We show that analyzing synthetic trajectories can support or contradict claims of identifiability. While we perform this on a specific model system, this approach can be generalized for a variety of stochastic representations of partially observable systems. We also suggest data manipulations intended to improve identifiability that might be applicable in many systems of interest.

A survey on agents applications in healthcare: Opportunities, challenges and trends

BACKGROUND AND OBJECTIVE

The agent abstraction is a powerful one, developed decades ago to represent crucial aspects of artificial intelligence research. The meaning has transformed over the years and now there are different nuances across research communities. At its core, an agent is an autonomous computational entity capable of sensing, acting, and capturing interactions with other agents and its environment. This review examines how agent-based techniques have been implemented and evaluated in a specific and very important domain, i.e. healthcare research.

METHODS

We survey key areas of agent-based research in healthcare, e.g. individual and collective behaviours, communicable and non-communicable diseases, and social epidemiology. We propose a systematic search and critical review of relevant recent works, introduced by an exploratory network analysis.

RESULTS

Network analysis enables to devise out 5 main research clusters, the most active authors, and 4 main research topics.

CONCLUSIONS

Our findings support discussion of some future directions for increasing the value of agent-based approaches in healthcare.

School Virus Infection Simulator for customizing school schedules during COVID-19

Even as the COVID-19 pandemic raged worldwide, schools strived to provide consistent education to their students. In such situations, schools require customized schedules that can address the health concerns and safety of the students to safely reopen and remain open. School schedules can be customized in many ways, and different approaches' impact on education and effectiveness in reducing infectious risks are different. To address this issue, we developed the School Virus Infection Simulation-Model (SVISM) for teachers and education policymakers. By taking into account the students' lesson schedules, classroom volume, air circulation rates in the classrooms, and infectability of the students, SVISM simulates the spread of infection at a school. We demonstrate the impact of several school schedules in self-contained and departmentalized classrooms and evaluate them in terms of the maximum number of students infected simultaneously, and the percentage of face-to-face lessons. The results show that the impact of increasing the classroom ventilation rate is not as stable as that of customizing school schedules. In addition, school schedules can differently impact the maximum number of students infected simultaneously, depending on whether classrooms are self-contained or departmentalized. We found that the maximum number of students infected simultaneously under a certain schedule with 50 percentage of face-to-face lessons in self-contained classrooms is higher than the maximum number of students infected simultaneously having schedules with a higher percentage of face-to-face lessons; this phenomenon was not found in departmentalized classrooms. These results show that the SVISM can help teachers and education policymakers plan school schedules appropriately to reduce the maximum number of students infected simultaneously, while also maintaining a certain rate of face-to-face lessons.

Short-term and long-term epidemiological impacts of sustained vector control in various dengue endemic settings: A modelling study

As the most widespread viral infection transmitted by the Aedes mosquitoes, dengue has been estimated to cause 51 million febrile disease cases globally each year. Although sustained vector control remains key to reducing the burden of dengue, current understanding of the key factors that explain the observed variation in the short- and long-term vector control effectiveness across different transmission settings remains limited. We used a detailed individual-based model to simulate dengue transmission with and without sustained vector control over a 30-year time frame, under different transmission scenarios. Vector control effectiveness was derived for different time windows within the 30-year intervention period. We then used the extreme gradient boosting algorithm to predict the effectiveness of vector control given the simulation parameters, and the resulting machine learning model was interpreted using Shapley Additive Explanations. According to our simulation outputs, dengue transmission would be nearly eliminated during the early stage of sustained and intensive vector control, but over time incidence would gradually bounce back to the pre-intervention level unless the intervention is implemented at a very high level of intensity. The time point at which intervention ceases to be effective is strongly influenced not only by the intensity of vector control, but also by the pre-intervention transmission intensity and the individual-level heterogeneity in biting risk. Moreover, the impact of many transmission model parameters on the intervention effectiveness is shown to be modified by the intensity of vector control, as well as to vary over time. Our study has identified some of the critical drivers for the difference in the time-varying effectiveness of sustained vector control across different dengue endemic settings, and the insights obtained will be useful to inform future model-based studies that seek to predict the impact of dengue vector control in their local contexts.

Sustained vector control remains key to reducing the global burden of dengue. However, current understanding of the main drivers for the differences in the time-varying epidemiological impact of dengue vector control across different transmission settings remains limited. We developed an agent-based model and showed that in the absence of a highly effective intervention technology that is able to eliminate dengue transmission even in an entirely susceptible population, a fixed level of reduction in the Aedes abundance would only cause temporary reduction in dengue incidence. Furthermore, the time point at which intervention ceases to be effective is strongly influenced not only by the intensity of vector control and the pre-intervention transmission intensity, but also by the individual-level heterogeneity in biting risk. Besides, the intensity of vector control interacts with the other two factors mentioned earlier, and the interaction magnitude also changes over time. These insights will be useful to inform future modelling studies that seek to predict the impact of Aedes control on dengue transmission in their local contexts, and have important implications for the design of intervention strategies to achieve sustained reduction in the global burden of dengue.

Dengue Vaccine: Recommendations of the Advisory Committee on Immunization Practices, United States, 2021

Dengue is a vectorborne infectious disease caused by dengue viruses (DENVs), which are predominantly transmitted by Aedes aegypti and Aedes albopictus mosquitos. Dengue is caused by four closely related viruses (DENV-1–4), and a person can be infected with each serotype for a total of four infections during their lifetime. Areas where dengue is endemic in the United States and its territories and freely associated states include Puerto Rico, American Samoa, the U.S. Virgin Islands, the Federated States of Micronesia, the Republic of Marshall Islands, and the Republic of Palau. This report summarizes the recommendations of the Advisory Committee on Immunization Practices (ACIP) for use of the Dengvaxia vaccine in the United States. The vaccine is a live-attenuated, chimeric tetravalent dengue vaccine built on a yellow fever 17D backbone. Dengvaxia is safe and effective in reducing dengue-related hospitalizations and severe dengue among persons who have had dengue infection in the past. Previous natural infection is important because Dengvaxia is associated with an increased risk for severe dengue in those who experience their first natural infection (i.e., primary infection) after vaccination. Dengvaxia was licensed by the Food and Drug Administration for use among children and adolescents aged 9–16 years (referred to in this report as children). ACIP recommends vaccination with Dengvaxia for children aged 9–16 having evidence of a previous dengue infection and living in areas where dengue is endemic. Evidence of previous dengue infection, such as detection of anti-DENV immunoglobulin G with a highly specific serodiagnostic test, will be required for eligible children before vaccination.

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.

Bridging implementation gaps to connect large ecological datasets and complex models

Merging robust statistical methods with complex simulation models is a frontier for improving ecological inference and forecasting. However, bringing these tools together is not always straightforward. Matching data with model output, determining starting conditions, and addressing high dimensionality are some of the complexities that arise when attempting to incorporate ecological field data with mechanistic models directly using sophisticated statistical methods. To illustrate these complexities and pragmatic paths forward, we present an analysis using tree-ring basal area reconstructions in Denali National Park (DNPP) to constrain successional trajectories of two spruce species (Picea mariana and Picea glauca) simulated by a forest gap model, University of Virginia Forest Model Enhanced-UVAFME. Through this process, we provide preliminary ecological inference about the long-term competitive dynamics between slow-growing P. mariana and relatively faster-growing P. glauca. Incorporating tree-ring data into UVAFME allowed us to estimate a bias correction for stand age with improved parameter estimates. We found that higher parameter values for P. mariana minimum growth under stress and P. glauca maximum growth rate were key to improving simulations of coexistence, agreeing with recent research that faster-growing P. glauca may outcompete P. mariana under climate change scenarios. The implementation challenges we highlight are a crucial part of the conversation for how to bring models together with data to improve ecological inference and forecasting.

Simulation Agent-Based Model to Demonstrate the Transmission of COVID-19 and Effectiveness of Different Public Health Strategies

COVID-19 has changed the world fundamentally since its outbreak in January 2020. Public health experts and administrations around the world suggested and implemented various intervention strategies to slow down the transmission of the virus. To illustrate to the general public how the virus is transmitted and how different intervention strategies can check the transmission, we built an agent-based model (ABM) to simulate the transmission of the virus in the real world and demonstrate how to prevent its spread with public health strategies.

Case-area targeted interventions (CATI) for reactive dengue control: Modelling effectiveness of vector control and prophylactic drugs in Singapore

BACKGROUND

Targeting interventions to areas that have recently experienced cases of disease is one strategy to contain outbreaks of infectious disease. Such case-area targeted interventions (CATI) have become an increasingly popular approach for dengue control but there is little evidence to suggest how precisely targeted or how recent cases need to be, to mount an effective response. The growing interest in the development of prophylactic and therapeutic drugs for dengue has also given new relevance for CATI strategies to interrupt transmission or deliver early treatment.

METHODS/PRINCIPAL FINDINGS

Here we develop a patch-based mathematical model of spatial dengue spread and fit it to spatiotemporal datasets from Singapore. Simulations from this model suggest CATI strategies could be effective, particularly if used in lower density areas. To maximise effectiveness, increasing the size of the radius around an index case should be prioritised even if it results in delays in the intervention being applied. This is partially because large intervention radii ensure individuals receive multiple and regular rounds of drug dosing or vector control, and thus boost overall coverage. Given equivalent efficacy, CATIs using prophylactic drugs are predicted to be more effective than adult mosquito-killing vector control methods and may even offer the possibility of interrupting individual chains of transmission if rapidly deployed. CATI strategies quickly lose their effectiveness if baseline transmission increases or case detection rates fall.

CONCLUSIONS/SIGNIFICANCE

These results suggest CATI strategies can play an important role in dengue control but are likely to be most relevant for low transmission areas where high coverage of other non-reactive interventions already exists. Controlled field trials are needed to assess the field efficacy and practical constraints of large operational CATI strategies.

Pandemic-associated mobility restrictions could cause increases in dengue virus transmission

BACKGROUND

The COVID-19 pandemic has induced unprecedented reductions in human mobility and social contacts throughout the world. Because dengue virus (DENV) transmission is strongly driven by human mobility, behavioral changes associated with the pandemic have been hypothesized to impact dengue incidence. By discouraging human contact, COVID-19 control measures have also disrupted dengue vector control interventions, the most effective of which require entry into homes. We sought to investigate how and why dengue incidence could differ under a lockdown scenario with a proportion of the population sheltered at home.

METHODOLOGY & PRINCIPAL FINDINGS

We used an agent-based model with a realistic treatment of human mobility and vector control. We found that a lockdown in which 70% of the population sheltered at home and which occurred in a season when a new serotype invaded could lead to a small average increase in cumulative DENV infections of up to 10%, depending on the time of year lockdown occurred. Lockdown had a more pronounced effect on the spatial distribution of DENV infections, with higher incidence under lockdown in regions with higher mosquito abundance. Transmission was also more focused in homes following lockdown. The proportion of people infected in their own home rose from 54% under normal conditions to 66% under lockdown, and the household secondary attack rate rose from 0.109 to 0.128, a 17% increase. When we considered that lockdown measures could disrupt regular, city-wide vector control campaigns, the increase in incidence was more pronounced than with lockdown alone, especially if lockdown occurred at the optimal time for vector control.

CONCLUSIONS & SIGNIFICANCE

Our results indicate that an unintended outcome of lockdown measures may be to adversely alter the epidemiology of dengue. This observation has important implications for an improved understanding of dengue epidemiology and effective application of dengue vector control. When coordinating public health responses during a syndemic, it is important to monitor multiple infections and understand that an intervention against one disease may exacerbate another.

Cost-effectiveness of dengue vaccination in Puerto Rico

An effective and widely used vaccine could reduce the burden of dengue virus (DENV) around the world. DENV is endemic in Puerto Rico, where the dengue vaccine CYD-TDV is currently under consideration as a control measure. CYD-TDV has demonstrated efficacy in clinical trials in vaccinees who had prior dengue virus infection. However, in vaccinees who had no prior dengue virus infection, the vaccine had a modestly elevated risk of hospitalization and severe disease. The WHO therefore recommended a strategy of pre-vaccination screening and vaccination of seropositive persons. To estimate the cost-effectiveness and benefits of this intervention (i.e., screening and vaccination of seropositive persons) in Puerto Rico, we simulated 10 years of the intervention in 9-year-olds using an agent-based model. Across the entire population, we found that 5.5% (4.6%-6.3%) of dengue hospitalizations could be averted. However, we also found that 0.057 (0.045-0.073) additional hospitalizations could occur for every 1,000 people in Puerto Rico due to DENV-naïve children who were vaccinated following a false-positive test results for prior exposure. The ratio of the averted hospitalizations among all vaccinees to additional hospitalizations among DENV-naïve vaccinees was estimated to be 19 (13-24). At a base case cost of vaccination of 382 USD, we found an incremental cost-effectiveness ratio of 122,000 USD per QALY gained. Our estimates can provide information for considerations to introduce the CYD-TDV vaccine in Puerto Rico.

Socio-demographic and health factors drive the epidemic progression and should guide vaccination strategies for best COVID-19 containment

We propose and study an epidemiological model on a social network that takes into account heterogeneity of the population and different vaccination strategies. In particular, we study how the COVID-19 epidemics evolves and how it is contained by different vaccination scenarios by taking into account data showing that older people, as well as individuals with comorbidities and poor metabolic health, and people coming from economically depressed areas with lower quality of life in general, are more likely to develop severe COVID-19 symptoms, and quicker loss of immunity and are therefore more prone to reinfection. Our results reveal that the structure and the spatial arrangement of subpopulations are important epidemiological determinants. In a healthier society the disease spreads more rapidly but the consequences are less disastrous as in a society with more prevalent chronic comorbidities. If individuals with poor health are segregated within one community, the epidemic outcome is less favorable. Moreover, we show that, contrary to currently widely adopted vaccination policies, prioritizing elderly and other higher-risk groups is beneficial only if the supply of vaccine is high. If, however, the vaccination availability is limited, and if the demographic distribution across the social network is homogeneous, better epidemic outcomes are achieved if healthy people are vaccinated first. Only when higher-risk groups are segregated, like in elderly homes, their prioritization will lead to lower COVID-19 related deaths. Accordingly, young and healthy individuals should view vaccine uptake as not only protecting them, but perhaps even more so protecting the more vulnerable socio-demographic groups.

Immune life history, vaccination, and the dynamics of SARS-CoV-2 over the next 5 years

The future trajectory of the coronavirus disease 2019 (COVID-19) pandemic hinges on the dynamics of adaptive immunity against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); however, salient features of the immune response elicited by natural infection or vaccination are still uncertain. We use simple epidemiological models to explore estimates for the magnitude and timing of future COVID-19 cases, given different assumptions regarding the protective efficacy and duration of the adaptive immune response to SARS-CoV-2, as well as its interaction with vaccines and nonpharmaceutical interventions. We find that variations in the immune response to primary SARS-CoV-2 infections and a potential vaccine can lead to markedly different immune landscapes and burdens of critically severe cases, ranging from sustained epidemics to near elimination. Our findings illustrate likely complexities in future COVID-19 dynamics and highlight the importance of immunological characterization beyond the measurement of active infections for adequately projecting the immune landscape generated by SARS-CoV-2 infections.

Hidden heterogeneity and its influence on dengue vaccination impact

The CYD-TDV vaccine was recently developed to combat dengue, a mosquito-borne viral disease that afflicts millions of people each year throughout the tropical and subtropical world. Its rollout has been complicated by recent findings that vaccinees with no prior exposure to dengue virus (DENV) experience an elevated risk of severe disease in response to their first DENV infection subsequent to vaccination. As a result of these findings, guidelines for use of CYD-TDV now require serological screening prior to vaccination to establish that an individual does not fall into this high-risk category. These complications mean that the public health impact of CYD-TDV vaccination is expected to be higher in areas with higher transmission. One important practical difficulty with tailoring vaccination policy to local transmission contexts is that DENV transmission is spatially heterogeneous, even at the scale of neighborhoods or blocks within a city. This raises the question of whether models based on data that average over spatial heterogeneity in transmission could fail to capture important aspects of CYD-TDV impact in spatially heterogeneous populations. We explored this question with a deterministic model of DENV transmission and CYD-TDV vaccination in a population comprised of two communities with differing transmission intensities. Compared to the full model, a version of the model based on the average of the two communities failed to capture benefits of targeting the intervention to the high-transmission community, which resulted in greater impact in both communities than we observed under even coverage. In addition, the model based on the average of the two communities substantially overestimated impact among vaccinated individuals in the low-transmission community. In the event that the specificity of serological screening is not high, this result suggests that models that ignore spatial heterogeneity could overlook the potential for harm to this segment of the population.

RC, Perkins TA. Optimizing the deployment of ultra-low volume and targeted indoor residual spraying for dengue outbreak response

Recent years have seen rising incidence of dengue and large outbreaks of Zika and chikungunya, which are all caused by viruses transmitted by Aedes aegypti mosquitoes. In most settings, the primary intervention against Aedes-transmitted viruses is vector control, such as indoor, ultra-low volume (ULV) spraying. Targeted indoor residual spraying (TIRS) has the potential to more effectively impact Aedes-borne diseases, but its implementation requires careful planning and evaluation. The optimal time to deploy these interventions and their relative epidemiological effects are, however, not well understood. We used an agent-based model of dengue virus transmission calibrated to data from Iquitos, Peru to assess the epidemiological effects of these interventions under differing strategies for deploying them. Specifically, we compared strategies where spray application was initiated when incidence rose above a threshold based on incidence in recent years to strategies where spraying occurred at the same time(s) each year. In the absence of spraying, the model predicted 361,000 infections [inter-quartile range (IQR): 347,000-383,000] in the period 2000-2010. The ULV strategy with the fewest median infections was spraying twice yearly, in March and October, which led to a median of 172,000 infections [IQR: 158,000-183,000], a 52% reduction from baseline. Compared to spraying once yearly in September, the best threshold-based strategy utilizing ULV had fewer median infections (254,000 vs. 261,000), but required more spraying (351 vs. 274 days). For TIRS, the best strategy was threshold-based, which led to the fewest infections of all strategies tested (9,900; [IQR: 8,720-11,400], a 94% reduction), and required fewer days spraying than the equivalent ULV strategy (280). Although spraying twice each year is likely to avert the most infections, our results indicate that a threshold-based strategy can become an alternative to better balance the translation of spraying effort into impact, particularly if used with a residual insecticide.

Temperature-dependent variation in the extrinsic incubation period elevates the risk of vector-borne disease emergence

Identifying ecological drivers of disease transmission is central to understanding disease risks. For vector-borne diseases, temperature is a major determinant of transmission because vital parameters determining the fitness of parasites and vectors are highly temperature-sensitive, including the extrinsic incubation period required for parasites to develop within the vector. Temperature also underlies dramatic differences in the individual-level variation in the extrinsic incubation period, yet the influence of this variation in disease transmission is largely unexplored. We incorporate empirical estimates of dengue virus extrinsic incubation period and its variation across a range of temperatures into a stochastic model to examine the consequences for disease emergence. We find that such variation impacts the probability of disease emergence because exceptionally rapid, but empirically observed incubation - typically ignored by modelling only the average - increases the chance of disease emergence even at the limits of the temperature range for dengue transmission. We show that variation in the extrinsic incubation period causes the greatest proportional increase in the risk of disease emergence at cooler temperatures where the mean incubation period is long, and associated variation is large. Thus, ignoring EIP variation will likely lead to underestimation of the risk of vector-borne disease emergence in temperate climates.