Prediction of measles cases in US counties: A machine learning approach.

BACKGROUND: Although measles was declared eliminated from the United States in 2000, the frequency of measles outbreaks has increased in recent years. The ability to predict the locations of future cases could aid efforts to prevent and contain measles in the United States. METHODS: We estimated county-level measles risk using a machine learning model with 17 predictor variables, which was trained on 2014 and 2018 United States county-level measles case data and tested on data from 2019. We compared the predicted and actual locations of 2019 measles cases. RESULTS: The model accurately predicted 95 % (specificity) of United States counties without measles cases and 72 % (sensitivity) of the United States counties that experienced ≥1 measles case in 2019, accounting for 94 % of all measles cases in 2019. Among the top 30 counties with the highest risk scores, the model accurately predicted 22 (73 %) counties with a measles case in 2019, corresponding to 72 % of all measles cases. CONCLUSIONS: This machine learning model accurately predicted a majority of the United States counties at high risk for measles and could be used as a framework by state and national health agencies in their measles prevention and containment efforts.

Adjusting mobile phone data to account for children's travel and the impact on measles dynamics in Zambia.

Models of measles transmission can be used to identify areas of high risk to tailor immunization strategies. Estimates of spatial connectivity can be derived from data such as mobile phone records, however it is not clear how this maps to the movement of children who are more likely to be infected. Using travel surveys across two districts in Zambia and national mobile phone data, we compared estimates of out-of-district travel for the population captured in the mobile phone data and child-specific travel from travel surveys. We then evaluated the impact of unadjusted and adjusted connectivity measures on simulated measles virus introduction events across Zambia. The number of trips made by children from the travel survey was three to five times lower than the general population estimates from mobile phone data. This decreased the percentage of districts with measles virus introduction events from 78% when using unadjusted data to 51% - 64% following adjustment. Failure to account for age-specific heterogeneities in travel estimated from mobile phone data resulted in overestimating subnational areas at high risk of introduction events, which could divert mitigation efforts to districts that are at lower risk.

Incorporating social determinants of health into transmission modeling of COVID-19 vaccine in the US: a scoping review.

During COVID-19 in the US, social determinants of health (SDH) have driven health disparities. However, the use of SDH in COVID-19 vaccine modeling is unclear. This review aimed to summarize the current landscape of incorporating SDH into COVID-19 vaccine transmission modeling in the US. Medline and Embase were searched up to October 2022. We included studies that used transmission modeling to assess the effects of COVID-19 vaccine strategies in the US. Studies' characteristics, factors incorporated into models, and approaches to incorporate these factors were extracted. Ninety-two studies were included. Of these, 11 studies incorporated SDH factors (alone or combined with demographic factors). Various sets of SDH factors were integrated, with occupation being the most common (8 studies), followed by geographical location (5 studies). The results show that few studies incorporate SDHs into their models, highlighting the need for research on SDH impact and approaches to incorporating SDH into modeling. Funding: This research was funded by the Centers for Disease Control and Prevention (CDC).

Understanding the impact of HIV on mpox transmission in the MSM population: A mathematical modeling study.

The recent mpox outbreak (in 2022-2023) has different clinical and epidemiological features compared with previous outbreaks of the disease. During this outbreak, sexual contact was believed to be the primary transmission route of the disease. In addition, the community of men having sex with men (MSM) was disproportionately affected by the outbreak. This population is also disproportionately affected by HIV infection. Given that both diseases can be transmitted sexually, the endemicity of HIV, and the high sexual behavior associated with the MSM community, it is essential to understand the effect of the two diseases spreading simultaneously in an MSM population. Particularly, we aim to understand the potential effects of HIV on an mpox outbreak in the MSM population. We develop a mechanistic mathematical model of HIV and mpox co-infection. Our model incorporates the dynamics of both diseases and considers HIV treatment with anti-retroviral therapy (ART). In addition, we consider a potential scenario where HIV infection increases susceptibility to mpox, and investigate the potential impact of this mechanism on mpox dynamics. Our analysis shows that HIV can facilitate the spread of mpox in an MSM population, and that HIV treatment with ART may not be sufficient to control the spread of mpox in the population. However, we showed that a moderate use of condoms or reduction in sexual contact in the population combined with ART is beneficial in controlling mpox transmission. Based on our analysis, it is evident that effective control of HIV, specifically through substantial ART use, moderate condom compliance, and reduction in sexual contact, is imperative for curtailing the transmission of mpox in an MSM population and mitigating the compounding impact of these intertwined epidemics.

A mathematical model for the within-host (re)infection dynamics of SARS-CoV-2.

Interactions between SARS-CoV-2 and the immune system during infection are complex. However, understanding the within-host SARS-CoV-2 dynamics is of enormous importance for clinical and public health outcomes. Current mathematical models focus on describing the within-host SARS-CoV-2 dynamics during the acute infection phase. Thereby they ignore important long-term post-acute infection effects. We present a mathematical model, which not only describes the SARS-CoV-2 infection dynamics during the acute infection phase, but extends current approaches by also recapitulating clinically observed long-term post-acute infection effects, such as the recovery of the number of susceptible epithelial cells to an initial pre-infection homeostatic level, a permanent and full clearance of the infection within the individual, immune waning, and the formation of long-term immune capacity levels after infection. Finally, we used our model and its description of the long-term post-acute infection dynamics to explore reinfection scenarios differentiating between distinct variant-specific properties of the reinfecting virus. Together, the model's ability to describe not only the acute but also the long-term post-acute infection dynamics provides a more realistic description of key outcomes and allows for its application in clinical and public health scenarios.

COVSIM: A stochastic agent-based COVID-19 SIMulation model for North Carolina.

We document the evolution and use of the stochastic agent-based COVID-19 simulation model (COVSIM) to study the impact of population behaviors and public health policy on disease spread within age, race/ethnicity, and urbanicity subpopulations in North Carolina. We detail the methodologies used to model the complexities of COVID-19, including multiple agent attributes (i.e., age, race/ethnicity, high-risk medical status), census tract-level interaction network, disease state network, agent behavior (i.e., masking, pharmaceutical intervention (PI) uptake, quarantine, mobility), and variants. We describe its uses outside of the COVID-19 Scenario Modeling Hub (CSMH), which has focused on the interplay of nonpharmaceutical and pharmaceutical interventions, equitability of vaccine distribution, and supporting local county decision-makers in North Carolina. This work has led to multiple publications and meetings with a variety of local stakeholders. When COVSIM joined the CSMH in January 2022, we found it was a sustainable way to support new COVID-19 challenges and allowed the group to focus on broader scientific questions. The CSMH has informed adaptions to our modeling approach, including redesigning our high-performance computing implementation.

Rapid initiation of antiretroviral therapy in Turkey: a modeling study.

Background: To effectively control the HIV epidemic and meet global targets, policymakers recommend the rapid initiation of antiretroviral therapy (ART). Our study aims to investigate the effect of rapid ART programs on individuals diagnosed with HIV, considering varying coverage and initiation days after diagnosis, and compare it to standard-of-care ART treatment in Turkey. Methods: We used a dynamic compartmental model to simulate the dynamics of HIV infection in Turkey. Rapid treatment, defined as initiation of ART within 7 days of diagnosis, was contrasted with standard-of-care treatment, which starts within 30 days of diagnosis. This study considered three coverage levels (10%, 50%, and 90%) and two rapid periods (7 and 14 days after diagnosis), comparing them to standard-of-care treatment in evaluating the number of HIV infections between 2020 and 2030. Results: Annual HIV incidence and prevalence for a 10-year period were obtained from model projections. In the absence of a rapid ART program, the model projected approximately 444,000 new HIV cases while the number of cases were reduced to 345,000 (22% reduction) with 90% of diagnosed cases included in the rapid ART program. Similarly, 10% and 50% rapid ART coverage has resulted in 3% and 13% reduction in HIV prevalence over a 10-year period. Conclusion: Rapid ART demonstrates the potential to mitigate the increasing HIV incidence in Turkey by reducing the number of infections. The benefit of the rapid ART program could be substantial when the coverage of the program reaches above a certain percentage of diagnosed population.

Examining the Influence of Imbalanced Social Contact Matrices in Epidemic Models.

Transmissible infections such as those caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) spread according to who contacts whom. Therefore, many epidemic models incorporate contact patterns through contact matrices. Contact matrices can be generated from social contact survey data. However, the resulting matrices are often imbalanced, such that the total number of contacts reported by group A with group B do not match those reported by group B with group A. We examined the theoretical influence of imbalanced contact matrices on the estimated basic reproduction number (R0). We then explored how imbalanced matrices may bias model-based epidemic projections using an illustrative simulation model of SARS-CoV-2 with 2 age groups (<15 and ≥15 years). Models with imbalanced matrices underestimated the initial spread of SARS-CoV-2, had later time to peak incidence, and had smaller peak incidence. Imbalanced matrices also influenced cumulative infections observed per age group, as well as the estimated impact of an age-specific vaccination strategy. Stratified transmission models that do not consider contact balancing may generate biased projections of epidemic trajectory and the impact of targeted public health interventions. Therefore, modeling studies should implement and report methods used to balance contact matrices for stratified transmission models.

A perturbative approach to the analysis of many-compartment models characterized by the presence of waning immunity.

The waning of immunity after recovery or vaccination is a major factor accounting for the severity and prolonged duration of an array of epidemics, ranging from COVID-19 to diphtheria and pertussis. To study the effectiveness of different immunity level-based vaccination schemes in mitigating the impact of waning immunity, we construct epidemiological models that mimic the latter's effect. The total susceptible population is divided into an arbitrarily large number of discrete compartments with varying levels of disease immunity. We then vaccinate various compartments within this framework, comparing the value of [Formula: see text] and the equilibria locations for our systems to determine an optimal immunization scheme under natural constraints. Relying on perturbative analysis, we establish a number of results concerning the location, existence, and uniqueness of the system's endemic equilibria, as well as results on disease-free equilibria. We use numerical techniques to supplement our analytical ones, applying our model to waning immunity dynamics in pertussis, among other diseases. Our analytical results are applicable to a wide range of systems composed of arbitrarily many ODEs.

Optimizing age specific strategies of vaccination for prevention of cytomegalovirus infection in the US using agent-based simulation.

BACKGROUND: There is an urgent need to develop a cytomegalovirus (CMV) vaccine as it remains the leading cause of birth defects in the United States. While several CMV vaccine candidates are currently in late-stage clinical trials, the most effective vaccination program remains an open research question. METHODS: To take into account the critical uncertainties when evaluating the vaccine impact on both vertical (congenital) and horizontal CMV transmissions, we developed a CMV agent-based model representative of the US population and contact network structures. RESULTS: We evaluated 648 vaccination scenarios under various assumptions of vaccination age, vaccine efficacy, protection duration, and vaccination coverage. The optimal age of vaccination under all scenarios is shown to be during early childhood. However, a relatively modest benefit was also seen with vaccination of females of reproduction age (around age of 25) assuming near universal coverage and long vaccine-mediated protection. CONCLUSIONS: This study highlights the important need for a pediatric vaccination program in mitigating CMV in the United States. Our model is poised to investigate further location-based vaccine effectiveness questions in future planning of both clinical trials as well as eventual program implementation.

Novel Approach for Identification of Basic and Effective Reproduction Numbers Illustrated with COVID-19.

This paper presents a novel numerical technique for the identification of effective and basic reproduction numbers, Re and R0, for long-term epidemics, using an inverse problem approach. The method is based on the direct integration of the SIR (Susceptible-Infectious-Removed) system of ordinary differential equations and the least-squares method. Simulations were conducted using official COVID-19 data for the United States and Canada, and for the states of Georgia, Texas, and Louisiana, for a period of two years and ten months. The results demonstrate the applicability of the method in simulating the dynamics of the epidemic and reveal an interesting relationship between the number of currently infectious individuals and the effective reproduction number, which is a useful tool for predicting the epidemic dynamics. For all conducted experiments, the results show that the local maximum (and minimum) values of the time-dependent effective reproduction number occur approximately three weeks before the local maximum (and minimum) values of the number of currently infectious individuals. This work provides a novel and efficient approach for the identification of time-dependent epidemics parameters.

Effect of Transmission and Vaccination on Time to Dominance of Emerging Viral Strains: A Simulation-Based Study.

We studied the effect of transmissibility and vaccination on the time required for an emerging strain of an existing virus to dominate in the infected population using a simulation-based experiment. The emergent strain is assumed to be completely resistant to the available vaccine. A stochastic version of a modified SIR model for emerging viral strains was developed to simulate surveillance data for infections. The proportion of emergent viral strain infections among the infected was modeled using a logistic curve and the time to dominance (TTD) was recorded for each simulation. A factorial experiment was implemented to compare the TTD values for different transmissibility coefficients, vaccination rates, and initial vaccination coverage. We discovered a non-linear relationship between TTD and the relative transmissibility of the emergent strain for populations with low vaccination coverage. Furthermore, higher vaccination coverage and high vaccination rates in the population yielded significantly lower TTD values. Vaccinating susceptible individuals against the current strain increases the susceptible pool of the emergent virus, which leads to the emergent strain spreading faster and requiring less time to dominate the infected population.

Modeling the impact of the Omicron infection wave in Germany.

In November 2021, the first infection with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variant of concern (VOC) B.1.1.529 ('Omicron') was reported in Germany, alongside global reports of reduced vaccine efficacy (VE) against infections with this variant. The potential threat posed by its rapid spread in Germany was, at the time, difficult to predict. We developed a variant-dependent population-averaged susceptible-exposed-infected-recovered infectious-disease model that included information about variant-specific and waning VEs based on empirical data available at the time. Compared to other approaches, our method aimed for minimal structural and computational complexity and therefore enabled us to respond to changes in the situation in a more agile manner while still being able to analyze the potential influence of (non-)pharmaceutical interventions (NPIs) on the emerging crisis. Thus, the model allowed us to estimate potential courses of upcoming infection waves in Germany, focusing on the corresponding burden on intensive care units (ICUs), the efficacy of contact reduction strategies, and the success of the booster vaccine rollout campaign. We expected a large cumulative number of infections with the VOC Omicron in Germany with ICU occupancy likely remaining below capacity, nevertheless, even without additional NPIs. The projected figures were in line with the actual Omicron waves that were subsequently observed in Germany with respective peaks occurring in mid-February and mid-March. Most surprisingly, our model showed that early, strict, and short contact reductions could have led to a strong 'rebound' effect with high incidences after the end of the respective NPIs, despite a potentially successful booster campaign. The results presented here informed legislation in Germany. The methodology developed in this study might be used to estimate the impact of future waves of COVID-19 or other infectious diseases.

Assessing the accuracy of California county level COVID-19 hospitalization forecasts to inform public policy decision making.

BACKGROUND: The COVID-19 pandemic has highlighted the role of infectious disease forecasting in informing public policy. However, significant barriers remain for effectively linking infectious disease forecasts to public health decision making, including a lack of model validation. Forecasting model performance and accuracy should be evaluated retrospectively to understand under which conditions models were reliable and could be improved in the future. METHODS: Using archived forecasts from the California Department of Public Health's California COVID Assessment Tool ( https://calcat.covid19.ca.gov/cacovidmodels/ ), we compared how well different forecasting models predicted COVID-19 hospitalization census across California counties and regions during periods of Alpha, Delta, and Omicron variant predominance. RESULTS: Based on mean absolute error estimates, forecasting models had variable performance across counties and through time. When accounting for model availability across counties and dates, some individual models performed consistently better than the ensemble model, but model rankings still differed across counties. Local transmission trends, variant prevalence, and county population size were informative predictors for determining which model performed best for a given county based on a random forest classification analysis. Overall, the ensemble model performed worse in less populous counties, in part because of fewer model contributors in these locations. CONCLUSIONS: Ensemble model predictions could be improved by incorporating geographic heterogeneity in model coverage and performance. Consistency in model reporting and improved model validation can strengthen the role of infectious disease forecasting in real-time public health decision making.

Predicting Measles Outbreaks in the United States: Evaluation of Machine Learning Approaches.

BACKGROUND: Measles, a highly contagious viral infection, is resurging in the United States, driven by international importation and declining domestic vaccination coverage. Despite this resurgence, measles outbreaks are still rare events that are difficult to predict. Improved methods to predict outbreaks at the county level would facilitate the optimal allocation of public health resources. OBJECTIVE: We aimed to validate and compare extreme gradient boosting (XGBoost) and logistic regression, 2 supervised learning approaches, to predict the US counties most likely to experience measles cases. We also aimed to assess the performance of hybrid versions of these models that incorporated additional predictors generated by 2 clustering algorithms, hierarchical density-based spatial clustering of applications with noise (HDBSCAN) and unsupervised random forest (uRF). METHODS: We constructed a supervised machine learning model based on XGBoost and unsupervised models based on HDBSCAN and uRF. The unsupervised models were used to investigate clustering patterns among counties with measles outbreaks; these clustering data were also incorporated into hybrid XGBoost models as additional input variables. The machine learning models were then compared to logistic regression models with and without input from the unsupervised models. RESULTS: Both HDBSCAN and uRF identified clusters that included a high percentage of counties with measles outbreaks. XGBoost and XGBoost hybrid models outperformed logistic regression and logistic regression hybrid models, with the area under the receiver operating curve values of 0.920-0.926 versus 0.900-0.908, the area under the precision-recall curve values of 0.522-0.532 versus 0.485-0.513, and F2 scores of 0.595-0.601 versus 0.385-0.426. Logistic regression or logistic regression hybrid models had higher sensitivity than XGBoost or XGBoost hybrid models (0.837-0.857 vs 0.704-0.735) but a lower positive predictive value (0.122-0.141 vs 0.340-0.367) and specificity (0.793-0.821 vs 0.952-0.958). The hybrid versions of the logistic regression and XGBoost models had slightly higher areas under the precision-recall curve, specificity, and positive predictive values than the respective models that did not include any unsupervised features. CONCLUSIONS: XGBoost provided more accurate predictions of measles cases at the county level compared with logistic regression. The threshold of prediction in this model can be adjusted to align with each county's resources, priorities, and risk for measles. While clustering pattern data from unsupervised machine learning approaches improved some aspects of model performance in this imbalanced data set, the optimal approach for the integration of such approaches with supervised machine learning models requires further investigation.

Projecting social contact matrices to populations stratified by binary attributes with known homophily.

Contact networks are heterogeneous. People with similar characteristics are more likely to interact, a phenomenon called assortative mixing or homophily. Empirical age-stratified social contact matrices have been derived by extensive survey work. We lack however similar empirical studies that provide social contact matrices for a population stratified by attributes beyond age, such as gender, sexual orientation, or ethnicity. Accounting for heterogeneities with respect to these attributes can have a profound effect on model dynamics. Here, we introduce a new method, which uses linear algebra and non-linear optimization, to expand a given contact matrix to populations stratified by binary attributes with a known level of homophily. Using a standard epidemiological model, we highlight the effect homophily can have on model dynamics, and conclude by briefly describing more complicated extensions. The available Python source code enables any modeler to account for the presence of homophily with respect to binary attributes in contact patterns, ultimately yielding more accurate predictive models.

The changing health impact of vaccines in the COVID-19 pandemic: A modeling study.

Much of the world's population had already been infected with COVID-19 by the time the Omicron variant emerged at the end of 2021, but the scale of the Omicron wave was larger than any that had come before or has happened since, and it left a global imprinting of immunity that changed the COVID-19 landscape. In this study, we simulate a South African population and demonstrate how population-level vaccine effectiveness and efficiency changed over the course of the first 2 years of the pandemic. We then introduce three hypothetical variants and evaluate the impact of vaccines with different properties. We find that variant-chasing vaccines have a narrow window of dominating pre-existing vaccines but that a variant-chasing vaccine strategy may have global utility, depending on the rate of spread from setting to setting. Next-generation vaccines might be able to overcome uncertainty in pace and degree of viral evolution.

An individual level infectious disease model in the presence of uncertainty from multiple, imperfect diagnostic tests.

Bayesian compartmental infectious disease models yield important inference on disease transmission by appropriately accounting for the dynamics and uncertainty of infection processes. In addition to estimating transition probabilities and reproductive numbers, these statistical models allow researchers to assess the probability of disease risk and quantify the effectiveness of interventions. These infectious disease models rely on data collected from all individuals classified as positive based on various diagnostic tests. In infectious disease testing, however, such procedures produce both false-positives and false-negatives at varying rates depending on the sensitivity and specificity of the diagnostic tests being used. We propose a novel Bayesian spatio-temporal infectious disease modeling framework that accounts for the additional uncertainty in the diagnostic testing and classification process that provides estimates of the important transmission dynamics of interest to researchers. The method is applied to data on the 2006 mumps epidemic in Iowa, in which over 6,000 suspected mumps cases were tested using a buccal or oral swab specimen, a urine specimen, and/or a blood specimen. Although all procedures are believed to have high specificities, the sensitivities can be low and vary depending on the timing of the test as well as the vaccination status of the individual being tested.

Modeling the impact of child vaccination (5-11 y) on overall COVID-19 related hospitalizations and mortality in a context of omicron variant predominance and different vaccination coverage paces in Brazil.

Background: Developing countries have experienced significant COVID-19 disease burden. With the emergence of new variants, particularly omicron, the disease burden in children has increased. When the first COVID-19 vaccine was approved for use in children aged 5-11 years of age, very few countries recommended vaccination due to limited risk-benefit evidence for vaccination of this population. In Brazil, ranking second in the global COVID-19 death toll, the childhood COVID-19 disease burden increased significantly in early 2022. This prompted a risk-benefit assessment of the introduction and scaling-up of COVID-19 vaccination of children. Methods: To estimate the potential impact of vaccinating children aged 5-11 years with mRNA-based COVID-19 vaccine in the context of omicron dominance, we developed a discrete-time SEIR-like model stratified in age groups, considering a three-month time horizon. We considered three scenarios: No vaccination, slow, and maximum vaccination paces. In each scenario, we estimated the potential reduction in total COVID-19 cases, hospitalizations, deaths, hospitalization costs, and potential years of life lost, considering the absence of vaccination as the base-case scenario. Findings: We estimated that vaccinating at a maximum pace could prevent, between mid-January and April 2022, about 26,000 COVID-19 hospitalizations, and 4200 deaths in all age groups; of which 5400 hospitalizations and 410 deaths in children aged 5-11 years. Continuing vaccination at a slow/current pace would prevent 1450 deaths and 9700 COVID-19 hospitalizations in all age groups in this same time period; of which 180 deaths and 2390 hospitalizations in children only. Interpretation: Maximum vaccination of children results in a significant reduction of COVID-19 hospitalizations and deaths and should be enforced in developing countries with significant disease incidence in children. Funding: This manuscript was funded by the Brazilian Council for Scientific and Technology Development (CNPq - Process # 402834/2020-8).