Search

Here, we introduce a novel framework for modelling the spatiotemporal dynamics of disease spread known as conditional logistic individual-level models (CL-ILM's). This framework alleviates much of the computational burden associated with traditional spatiotemporal individual-level models for epidemics, and facilitates the use of standard software for fitting logistic models when analysing spatiotemporal disease patterns. The models can be fitted in either a frequentist or Bayesian framework. Here, we apply the new spatial CL-ILM to simulated data, semi-real data from the UK 2001 foot-and-mouth disease epidemic, and real data from a greenhouse experiment on the spread of tomato spotted wilt virus.

Human behaviour significantly affects the dynamics of infectious disease transmission as people adjust their behavior in response to outbreak intensity, thereby impacting disease spread and control efforts. In recent years, there have been efforts to incorporate behavioural change into spatio-temporal individual-level models within a Bayesian MCMC framework. In this past work, parametric spatial risk functions were employed, depending on strong underlying assumptions regarding disease transmission mechanisms within the population. However, selecting appropriate parametric functions can be challenging in real-world scenarios, and incorrect assumptions may lead to erroneous conclusions. As an alternative, non-parametric approaches offer greater flexibility. The goal of this study is to investigate the utilization of semi-parametric spatial models for infectious disease transmission, integrating an “alarm function” to account for behavioural change based on infection prevalence over time within a Bayesian MCMC framework. In this paper, we discuss findings from both simulated and real-life epidemics, focusing on constant piecewise distance functions with fixed change points. We also demonstrate the selection of the change points using the Deviance Information Criteria (DIC).

ABSTRACT: Mycoplasma bovis outbreaks in cattle, including pathogen spread between age groups, are not well understood. Our objective was to estimate within-herd transmission across adult dairy cows, youngstock, and calves. Results from 3 tests (PCR, ELISA, and culture) per cow and 2 tests (PCR and ELISA) per youngstock and calf were used in an age-stratified susceptible-infected-removed/recovered (SIR) model to estimate within-herd transmission parameters, pathways, and potential effects of farm management practices. A cohort of adult cows, youngstock, and calves on 20 Dutch dairy farms with a clinical outbreak of M. bovis in adult cows were sampled, with collection of blood, conjunctival fluid, and milk from cows, and blood and conjunctival fluid from calves and youngstock, 5 times over a time span of 12 wk. Any individual with at least one positive laboratory test was considered M. bovis-positive. Transmission dynamics were modeled using an age-stratified SIR model featuring 3 age strata. Associations with farm management practices were explored using Fisher's exact tests and Poisson regression. Estimated transmission parameters were highly variable among herds and cattle age groups. Notably, transmission from cows to cows, youngstock, or to calves was associated with R-values ranging from 1.0 to 80 secondarily infected cows per herd, 1.2 to 38 secondarily infected youngstock per herd, and 0.1 to 91 secondarily infected calves per herd, respectively. In case of transmission from youngstock to youngstock, calves or to cows, R-values were 0.7 to 96 secondarily infected youngstock per herd, 1.1 to 76 secondarily infected calves per herd, and 0.1 to 107 secondarily infected cows per herd. For transmission from calves to calves, youngstock or to cows, R-values were 0.5 to 60 secondarily infected calves per herd, 1.1 to 41 secondarily infected youngstock per herd, and 0.1 to 47 secondarily infected cows per herd. Among on-farm transmission pathways, cow-to-youngstock, cow-to-calf, and cow-to-cow were identified as most significant contributors, with calf-to-calf and calf-to-youngstock also having noteworthy roles. Youngstock-to-youngstock was also implicated, albeit to a lesser extent. Whereas the primary focus was a clinical outbreak of M. bovis among adult dairy cows, it was evident that transmission extended to calves and youngstock, contributing to overall spread. Factors influencing transmission and specific transmission pathways were associated with internal biosecurity (separate caretakers for various age groups, number of people involved), external biosecurity (contractors, external employees), as well as indirect transmission routes (number of feed and water stations).

For many infectious disease outbreaks, the at-risk population changes their behavior in response to the outbreak severity, causing the transmission dynamics to change in real-time. Behavioral change is often ignored in epidemic modeling efforts, making these models less useful than they could be. We address this by introducing a novel class of data-driven epidemic models which characterize and accurately estimate behavioral change. Our proposed model allows time-varying transmission to be captured by the level of “alarm” in the population, with alarm specified as a function of the past epidemic trajectory. We investigate the estimability of the population alarm across a wide range of scenarios, applying both parametric functions and non-parametric functions using splines and Gaussian processes. The model is set in the data-augmented Bayesian framework to allow estimation on partially observed epidemic data. The benefit and utility of the proposed approach is illustrated through applications to data from real epidemics.

We introduce Pathogen.jl for simulation and inference of transmission network individual level models (TN-ILMs) of infectious disease spread in continuous time. TN-ILMs can be used to jointly infer transmission networks, event times, and model parameters within a Bayesian framework via Markov chain Monte Carlo (MCMC). We detail our specific strategies for conducting MCMC for TN-ILMs, and our implementation of these strategies in the Julia package, Pathogen.jl, which leverages key features of the Julia language. We provide an example using Pathogen.jl to simulate an epidemic following a susceptible-infectious-removed (SIR) TN-ILM, and then perform inference using observations that were generated from that epidemic. We also demonstrate the functionality of Pathogen.jl with an application of TN-ILMs to data from a measles outbreak that occurred in Hagelloch, Germany, in 1861 (Pfeilsticker 1863; Oesterle 1992).

Management of coronavirus disease 2019 (COVID-19) in India is a top government priority.However, there is a lack of COVID-19 adjusted case fatality risk (aCFR) estimates and information on states with high aCFR. Data on COVID-19 cases and deaths in the first pandemic wave and 17 state-specific geodemographic, socio-economic, health and comorbidity-related factors were collected. State-specific aCFRs were estimated, using a 13-day lag for fatality. To estimate country-level aCFR in the first wave, state estimates were meta-analysed based on inverse-variance weighting and aCFR as either a fixed- or random-effect. Multiple correspondence analyses, followed by univariable logistic regression, were conducted to understand the association between aCFR and geodemographic, health and social indicators.Based on health indicators, states likely to report a higher aCFR were identified. Using random- and fixed-effects models, cumulative aCFRs in the first pandemic wave on 27 July 2020 in India were 1.42% (95% CI 1.19%–1.70%) and 2.97% (95% CI 2.94%–3.00%), respectively. At the end of the first wave, as of 15 February 2021, a cumulative aCFR of 1.18% (95% CI 0.99%–1.41%) using random and 1.64% (95% CI 1.64%–1.65%) using fixed-effects models was estimated. Based on high heterogeneity among states, we inferred that the random-effects model likely provided more accurate estimates of the aCFR for India. The aCFR was grouped with the incidence of diabetes, hypertension, cardiovascular diseases and acute respiratory infections in the first and second dimensions of multiple correspondence analyses. Univariable logistic regression confirmed associations between the aCFR and the proportion of urban population, and between aCFR and the number of persons diagnosed with diabetes, hypertension, cardiovascular diseases and stroke per 10,000 population that had visited NCD (Non-communicable disease) clinics. Incidence of pneumonia was also associated with COVID-19 aCFR. Based on predictor variables, we categorised 10, 17 and one Indian state(s) expected to have a high, medium and low aCFR risk, respectively. The current study demonstrated the value of using meta-analysis to estimate aCFR. To decrease COVID-19 associated fatalities, states estimated to have a high aCFR must take steps to reduce co-morbidities.

A broad, cross-sectional study of beef cattle at entry into Canadian feedlots investigated the prevalence and epidemiology of antimicrobial resistance (AMR) in Mannheimia haemolytica, Pasteurella multocida, Histophilus somni, and Mycoplasma bovis, bacterial members of the bovine respiratory disease (BRD) complex. Upon feedlot arrival and before antimicrobials were administered at the feedlot, deep nasopharyngeal swabs were collected from 2,824 feedlot cattle in southern and central Alberta, Canada. Data on the date of feedlot arrival, cattle type (beef, dairy), sex (heifer, bull, steer), weight (kg), age class (calf, yearling), source (ranch direct, auction barn, backgrounding operations), risk of developing BRD (high, low), and weather conditions at arrival (temperature, precipitation, and estimated wind speed) were obtained. Mannheimia haemolytica, P. multocida, and H. somni isolates with multidrug-resistant (MDR) profiles associated with the presence of integrative and conjugative elements were isolated more often from dairy-type than from beef-type cattle. Our results showed that beef-type cattle from backgrounding operations presented higher odds of AMR bacteria as compared to auction-derived calves. Oxytetracycline resistance was the most frequently observed resistance across all Pasteurellaceae species and cattle types. Mycoplasma bovis exhibited high macrolide minimum inhibitory concentrations in both cattle types. Whether these MDR isolates establish and persist within the feedlot environment, requires further evaluation.

Objective To jointly estimate American College of Rheumatology (ACR50) response (a more commonly reported outcome) and remission (a more clinically relevant outcome) for methotrexate (MTX)‐based treatment options in rheumatoid arthritis (RA). Methods We conducted a bivariate network meta‐analysis (NMA) to compare MTX monotherapy and MTX‐based conventional and biologic disease‐modifying antirheumatic drug (DMARD) combinations for RA. The correlation between the outcomes was derived from an incident RA cohort study, whereas the treatment effects were derived from randomized trials in the network of evidence. The analyses were conducted separately for MTX‐naïve and MTX–inadequate response (IR) populations in a Bayesian framework with uninformative priors. Results From the cohort study, the correlation between ACR50 response and Disease Activity Score 28 remission at 6 months was moderate (Pearson correlation coefficient = 0.58). In the bivariate NMA for MTX‐naïve populations, most combinations of MTX with either biologic or tofacitinib were statistically superior to MTX alone for both ACR50 response and remission. Triple therapy (MTX + sulfasalazine + hydroxychloroquine) was the only nonbiologic DMARD statistically superior to MTX for either ACR50 response (odds ratio [OR] 95% credible interval: 2.1 [1.0, 4.3]) or remission (OR: 2.5 [1.0, 5.8]). In the MTX‐IR analysis, all treatments except MTX + sulfasalazine were statistically superior to MTX alone. Compared to analyzing the outcomes separately, the bivariate model often resulted in more precise estimates and allowed remission to be estimated for all treatments. Conclusion Borrowing the strength of correlation between outcomes allowed us to demonstrate a statistically significant benefit for remission across most MTX‐based DMARD combinations, including triple therapy.

Abstract Background School absenteeism data have been collected daily by the public health unit in Wellington-Dufferin-Guelph, Ontario since 2008. To date, a threshold-based approach has been implemented to raise alerts for community-wide and within-school illness outbreaks. We investigate several statistical modelling approaches to using school absenteeism for influenza surveillance at the regional level, and compare their performances using two metrics. Methods Daily absenteeism percentages from elementary and secondary schools, and report dates for influenza cases, were obtained from Wellington-Dufferin-Guelph Public Health. Several absenteeism data aggregations were explored, including using the average across all schools or only using schools of one type. A 10% absence threshold, exponentially weighted moving average model, logistic regression with and without seasonality terms, day of week indicators, and random intercepts for school year, and generalized estimating equations were used as epidemic detection methods for seasonal influenza. In the regression models, absenteeism data with various lags were used as predictor variables, and missing values in the datasets used for parameter estimation were handled either by deletion or linear interpolation. The epidemic detection methods were compared using a false alarm rate (FAR) as well as a metric for alarm timeliness. Results All model-based epidemic detection methods were found to decrease the FAR when compared to the 10% absence threshold. Regression models outperformed the exponentially weighted moving average model and including seasonality terms and a random intercept for school year generally resulted in fewer false alarms. The best-performing model, a seasonal logistic regression model with random intercept for school year and a day of week indicator where parameters were estimated using absenteeism data that had missing values linearly interpolated, produced a FAR of 0.299, compared to the pre-existing threshold method which at best gave a FAR of 0.827. Conclusions School absenteeism can be a useful tool for alerting public health to upcoming influenza epidemics in Wellington-Dufferin-Guelph. Logistic regression with seasonality terms and a random intercept for school year was effective at maximizing true alarms while minimizing false alarms on historical data from this region.

This paper describes the R package EpiILMCT, which allows users to study the spread of infectious disease using continuous time individual level models (ILMs). The package provides tools for simulation from continuous time ILMs that are based on either spatial demographic, contact network, or a combination of both of them, and for the graphical summarization of epidemics. Model fitting is carried out within a Bayesian Markov Chain Monte Carlo framework. The continuous time ILMs can be implemented within either susceptible-infected-removed (SIR) or susceptible-infected-notified-removed (SIN R) compartmental frameworks. As infectious disease data is often partially observed, data uncertainties in the form of missing infection times - and in some situations missing removal times - are accounted for using data augmentation techniques. The package is illustrated using both simulated and an experimental data set on the spread of the tomato spotted wilt virus disease.

The ongoing COVID-19 pandemic has led to a serious health crisis, and information obtained from disease transmission models fitted to observed data is needed to inform containment strategies. As the transmission of virus varies from city to city in different countries, we use a two-level individual-level model to analyze the spatiotemporal SARS-CoV-2 spread. However, inference procedures such as Bayesian Markov chain Monte Carlo, which is commonly used to estimate parameters of ILMs, are computationally expensive. In this study, we use trained ensemble learning classifiers to estimate the parameters of two-level ILMs and show that the fitted ILMs can successfully capture the virus transmission among Wuhan and 16 other cities in Hubei province, China.

Accurate and reliable short-term forecasts of influenza-like illness (ILI) visit volumes at emergency departments can improve staffing and resource allocation decisions within hospitals. In this paper, we developed a stacked ensemble model that averages the predictions from various competing methodologies in the current frontier for ILI-related forecasts. We also constructed a back-of-the-envelope prediction interval for the stacked ensemble, which provides a conservative characterization of the uncertainty in the stacked ensemble predictions. We assessed the accuracy and reliability of our model with 1 to 4 weeks ahead forecast targets using real-time hospital-level data on weekly ILI visit volumes during the 2012-2018 flu seasons in the Alberta Children's Hospital, located in Calgary, Alberta, Canada. Our results suggest the forecasting performance of the stacked ensemble meets or exceeds the performance of the individual models over all forecast targets.

Abstract “To consult the statistician after an experiment is finished is often merely to ask him to conduct a post mortem examination. He can perhaps say what the experiment died of.” – R.A. Fisher While this idea is relevant across research scales, its importance becomes critical when dealing with the inherently large, complex and expensive process of preparing material for cell-based therapies (CBTs). Effective and economically viable CBTs will depend on the establishment of optimized protocols for the production of the necessary cell types. Our ability to do this will depend in turn on the capacity to efficiently search through a multi-dimensional problem space of possible protocols in a timely and cost-effective manner. In this review we discuss approaches to, and illustrate examples of the application of statistical design of experiments to stem cell bioprocess optimization.

Influenza A virus commonly circulating in swine (IAV-S) is characterized by large genetic and antigenic diversity and, thus, improvements in different aspects of IAV-S surveillance are needed to achieve desirable goals of surveillance such as to establish the capacity to forecast with the greatest accuracy the number of influenza cases likely to arise. Advancements in modeling approaches provide the opportunity to use different models for surveillance. However, in order to make improvements in surveillance, it is necessary to assess the predictive ability of such models. This study compares the sensitivity and predictive accuracy of the autoregressive integrated moving average (ARIMA) model, the generalized linear autoregressive moving average (GLARMA) model, and the random forest (RF) model with respect to the frequency of influenza A virus (IAV) in Ontario swine. Diagnostic data on IAV submissions in Ontario swine between 2007 and 2015 were obtained from the Animal Health Laboratory (University of Guelph, Guelph, ON, Canada). Each modeling approach was examined for predictive accuracy, evaluated by the root mean square error, the normalized root mean square error, and the model's ability to anticipate increases and decreases in disease frequency. Likewise, we verified the magnitude of improvement offered by the ARIMA, GLARMA and RF models over a seasonal-naïve method. Using the diagnostic submissions, the occurrence of seasonality and the long-term trend in IAV infections were also investigated. The RF model had the smallest root mean square error in the prospective analysis and tended to predict increases in the number of diagnostic submissions and positive virological submissions at weekly and monthly intervals with a higher degree of sensitivity than the ARIMA and GLARMA models. The number of weekly positive virological submissions is significantly higher in the fall calendar season compared to the summer calendar season. Positive counts at weekly and monthly intervals demonstrated a significant increasing trend. Overall, this study shows that the RF model offers enhanced prediction ability over the ARIMA and GLARMA time series models for predicting the frequency of IAV infections in diagnostic submissions.

A class of discrete-time models of infectious disease spread, referred to as individual-level models (ILMs), are typically fitted in a Bayesian Markov chain Monte Carlo (MCMC) framework. These models quantify probabilistic outcomes regarding the risk of infection of susceptible individuals due to various susceptibility and transmissibility factors, including their spatial distance from infectious individuals. The infectious pressure from infected individuals exerted on susceptible individuals is intrinsic to these ILMs. Unfortunately, quantifying this infectious pressure for data sets containing many individuals can be computationally burdensome, leading to a time-consuming likelihood calculation and, thus, computationally prohibitive MCMC-based analysis. This problem worsens when using data augmentation to allow for uncertainty in infection times. In this paper, we develop sampling methods that can be used to calculate a fast, approximate likelihood when fitting such disease models. A simple random sampling approach is initially considered followed by various spatially-stratified schemes. We test and compare the performance of our methods with both simulated data and data from the 2001 foot-and-mouth disease (FMD) epidemic in the U.K. Our results indicate that substantial computation savings can be obtained--albeit, of course, with some information loss--suggesting that such techniques may be of use in the analysis of very large epidemic data sets.

Since 2001 models of the spread of foot-and-mouth disease, supported by the data from the UK epidemic, have been expounded as some of the best examples of problem-driven epidemic models. These claims are generally based on a comparison between model results and epidemic data at fairly coarse spatio-temporal resolution. Here, we focus on a comparison between model and data at the individual farm level, assessing the potential of the model to predict the infectious status of farms in both the short and long terms. Although the accuracy with which the model predicts farms reporting infection is between 5 and 15%, these low levels are attributable to the expected level of variation between epidemics, and are comparable to the agreement between two independent model simulations. By contrast, while the accuracy of predicting culls is higher (20–30%), this is lower than expected from the comparison between model epidemics. These results generally support the contention that the type of the model used in 2001 was a reliable representation of the epidemic process, but highlight the difficulties of predicting the complex human response, in terms of control strategies to the perceived epidemic risk. [ABSTRACT FROM AUTHOR]