Risk factors associated with the intensity of COVID-19 outbreaks in Canadian community settings: a retrospective analysis of outbreak-level surveillance data.

BACKGROUND: The severity of COVID-19 outbreaks is disproportionate across settings (e.g., long-term care facilities (LTCF), schools) across Canada. Few studies have examined factors associated with outbreak severity to inform prevention and response. Our study objective was to assess how outbreak severity, as measured using outbreak intensity and defined as number of outbreak-associated cases divided by outbreak duration, differed by setting and factors known to influence SARS-CoV-2 transmission. METHODS: We described outbreak intensity trends in 2021 using data from the Canadian COVID-19 Outbreak Surveillance System from seven provinces/territories, representing 93% of the Canadian population. A negative binomial fixed-effects model was used to assess for associations between the outcome, outbreak intensity, and characteristics of outbreaks: setting type, median age of cases, number at risk, and vaccination coverage of at least 1 dose. Also included were variables previously reported to influence SARS-CoV-2 transmission: stringency of non-pharmaceutical interventions (NPI) and the predominant SARS-CoV-2 variant detected by surveillance. RESULTS: The longest outbreaks occurred in LTCF (mean = 25.4 days) and correctional facilities (mean = 20.6 days) which also reported the largest outbreaks (mean = 29.6 cases per outbreak). Model results indicated that outbreak intensity was highest in correctional facilities. Relative to correctional facilities (referent), the second highest adjusted intensity ratio was in childcare centres (intensity ratio = 0.58 [95% CI: 0.51-0.66]), followed by LTCF (0.56 [95% CI: 0.51-0.66]). Schools had the lowest adjusted intensity ratio (0.46 [95% CI: 0.40-0.53]) despite having the highest proportion of outbreaks (37.5%). An increase in outbreak intensity was associated with increases in median age, the number at risk, and stringency of NPI. Greater vaccination coverage with at least 1 dose was associated with reduced outbreak intensity. CONCLUSION: Descriptive and multivariable model results indicated that in Canada during 2021, outbreak intensity was greatest in closed congregate living facilities: correctional facilities and LTCF. Findings from this study support the importance of vaccination in reducing outbreak intensity when vaccines are effective against infection with circulating variants, which is especially important for closed congregate living facilities where NPIs are more challenging to implement.

Linking Opinions Shared on Social Media About COVID-19 Public Health Measures to Adherence: Repeated Cross-Sectional Surveys of Twitter Use in Canada.

BACKGROUND: The effectiveness of public health measures (PHMs) depends on population adherence. Social media were suggested as a tool to assess adherence, but representativeness and accuracy issues have been raised. OBJECTIVE: The objectives of this repeated cross-sectional study were to compare self-reported PHM adherence and sociodemographic characteristics between people who used Twitter (subsequently rebranded X) and people who did not use Twitter. METHODS: Repeated Canada-wide web-based surveys were conducted every 14 days from September 2020 to March 2022. Weighted proportions were calculated for descriptive variables. Using Bayesian logistic regression models, we investigated associations between Twitter use, as well as opinions in tweets, and self-reported adherence with mask wearing and vaccination. RESULTS: Data from 40,230 respondents were analyzed. As self-reported, Twitter was used by 20.6% (95% CI 20.1%-21.2%) of Canadians, of whom 29.9% (95% CI 28.6%-31.3%) tweeted about COVID-19. The sociodemographic characteristics differed across categories of Twitter use and opinions. Overall, 11% (95% CI 10.6%-11.3%) of Canadians reported poor adherence to mask-wearing, and 10.8% (95% CI 10.4%-11.2%) to vaccination. Twitter users who tweeted about COVID-19 reported poorer adherence to mask wearing than nonusers, which was modified by the age of the respondents and their geographical region (odds ratio [OR] 0.79, 95% Bayesian credibility interval [BCI] 0.18-1.69 to OR 4.83, 95% BCI 3.13-6.86). The odds of poor adherence to vaccination of Twitter users who tweeted about COVID-19 were greater than those of nonusers (OR 1.76, 95% BCI 1.48-2.07). English- and French-speaking Twitter users who tweeted critically of PHMs were more likely (OR 4.07, 95% BCI 3.38-4.80 and OR 7.31, 95% BCI 4.26-11.03, respectively) to report poor adherence to mask wearing than non-Twitter users, and those who tweeted in support were less likely (OR 0.47, 95% BCI 0.31-0.64 and OR 0.96, 95% BCI 0.18-2.33, respectively) to report poor adherence to mask wearing than non-Twitter users. The OR of poor adherence to vaccination for those tweeting critically about PHMs and for those tweeting in support of PHMs were 4.10 (95% BCI 3.40-4.85) and 0.20 (95% BCI 0.10-0.32), respectively, compared to non-Twitter users. CONCLUSIONS: Opinions shared on Twitter can be useful to public health authorities, as they are associated with adherence to PHMs. However, the sociodemographics of social media users do not represent the general population, calling for caution when using tweets to assess general population-level behaviors.

Sensitivity and specificity of International Classification of Diseases algorithms (ICD-9 and ICD-10) used to identify opioid-related overdose cases: A systematic review and an example of estimation using Bayesian latent class models in the absence of gold standards.

OBJECTIVES: This study aimed to summarize validity estimates of International Classification of Diseases (ICD) codes in identifying opioid overdose (OOD) among patient data from emergency rooms, emergency medical services, inpatient, outpatient, administrative, medical claims, and mortality, and estimate the sensitivity and specificity of the algorithms in the absence of a perfect reference standard. METHODS: We systematically reviewed studies published before December 8, 2023, and identified with Medline and Embase. Studies reporting sufficient details to recreate a 2 × 2 table comparing the ICD algorithms to a reference standard in diagnosing OOD-related events were included. We used Bayesian latent class models (BLCM) to estimate the posterior sensitivity and specificity distributions of five ICD-10 algorithms and of the imperfect coroner's report review (CRR) in detecting prescription opioid-related deaths (POD) using one included study. RESULTS: Of a total of 1990 studies reviewed, three were included. The reported sensitivity estimates of ICD algorithms for OOD were low (range from 25.0% to 56.8%) for ICD-9 in diagnosing non-fatal OOD-related events and moderate (72% to 89%) for ICD-10 in diagnosing POD. The last included study used ICD-9 for non-fatal and fatal and ICD-10 for fatal OOD-related events and showed high sensitivity (i.e. above 97%). The specificity estimates of ICD algorithms were good to excellent in the three included studies. The misclassification-adjusted ICD-10 algorithm sensitivity estimates for POD from BLCM were consistently higher than reported sensitivity estimates that assumed CRR was perfect. CONCLUSION: Evidence on the performance of ICD algorithms in detecting OOD events is scarce, and the absence of bias correction for imperfect tests leads to an underestimation of the sensitivity of ICD code estimates.

RéSUMé: OBJECTIFS: Cette étude avait pour objectifs de recenser les estimations de la validité des codes de Classification Internationale des Maladies (CIM) à diagnostiquer les cas de surdose aux opioïdes (SDO) chez des patients en utilisant les données de salles d'urgence, services médicaux d'urgence, hospitalisations, soins ambulatoires, services administratifs, demandes de remboursement de frais médicaux, ainsi que de mortalité, et d'estimer la sensibilité et la spécificité d'algorithmes utilisant la CIM en l'absence d'un test de référence parfait. MéTHODES: Nous avons examiné systématiquement les études publiées avant le 8 décembre 2023, et identifiées dans Medline et Embase. Les études rapportant suffisamment de détails permettant de recréer un tableau 2 × 2 comparant les algorithmes de la CIM à un test de référence pour le diagnostic d'événements liés aux SDO ont été incluses. Les données d'une étude éligible ont été utilisées pour estimer, avec des modèles Bayésiens de classes latentes (MBCL), les distributions a posteriori de la sensibilité et de la spécificité de cinq algorithmes de la CIM-10 et du test imparfait de révision du rapport du coroner (RRC) dans la détection des décès liés aux opioïdes de prescription (DOP). RéSULTATS: Trois parmi les 1 990 études examinées ont été retenues. Les estimations rapportées de la sensibilité des codes CIM étaient faibles (variant de 25,0 % à 56,8 %) pour CIM-9 dans le diagnostic des événements liés aux SDO non-fatales dans une étude, et modérées (72 % à 89 %) pour CIM-10 dans le diagnostic des DOP dans une autre étude. La dernière étude incluse combinait des codes CIM-9 pour les cas non-fatals et fatals et CIM-10 pour les cas fatals et démontrait des estimations de sensibilité élevées (c.à.d. supérieures à 97 %). Les estimations de la spécificité étaient bonnes à excellentes dans les trois études. Les estimations de la sensibilité des algorithmes de la CIM-10 corrigées pour les erreurs de classification pour les décès liés aux opioïdes, obtenues à partir de nos MBCL, étaient systématiquement plus élevées que celles rapportées et qui supposaient que RRC était un test parfait. CONCLUSION: Les évidences sur la performance des algorithmes de la CIM dans la détection des cas de SDO sont rares, et l'absence de correction de biais pour des tests diagnostiques imparfaits conduit à une sous-estimation de la sensibilité des codes de la CIM.

Spatiotemporal epidemiology of substance-related accidental acute toxicity deaths in Canada from 2016 to 2017.

OBJECTIVES: In Canada, substance-related accidental acute toxicity deaths (AATDs) continue to rise at the national and sub-national levels. However, it is unknown if, where, when, and to what degree AATDs cluster in space, time, and space-time across the country. The objectives of this study were to 1) assess for clusters of AATDs that occurred in Canada during 2016 and 2017 at the national and provincial/territorial (P/T) levels, and 2) examine the substance types detected in AATD cases within each cluster. METHODS: Two years of person-level data on AATDs were abstracted from coroner and medical examiner files using a standardized data collection tool, including the decedent's postal code and municipality information on the places of residence, acute toxicity (AT) event, and death, and the substances detected in the death. Data were combined with Canadian census information to create choropleth maps depicting AATD rates by census division. Spatial scan statistics were used to build Poisson models to identify clusters of high rates (p < 0.05) of AATDs at the national and P/T levels in space, time, and space-time over the study period. AATD cases within clusters were further examined for substance types most present in each cluster. RESULTS: Eight clusters in five regions of Canada at the national level and 24 clusters in 15 regions at the P/T level were identified, highlighting where AATDs occurred at far higher rates than the rest of the country. The risk ratios of identified clusters ranged from 1.28 to 9.62. Substances detected in clusters varied by region and time, however, opioids, stimulants, and alcohol were typically the most commonly detected substances within clusters. CONCLUSION: Our findings are the first in Canada to reveal the geographic disparities in AATDs at national and P/T levels using spatial scan statistics. Rates associated with substance types within each cluster highlight which substance types were most detected in the identified regions. Findings may be used to guide intervention/program planning and provide a picture of the 2016 and 2017 context that can be used for comparisons of the geographic distribution of AATDs and substances with different time periods.

Impact of the COVID-19 pandemic on inbound air travel to Canada.

Background: Commercial air travel can result in global dispersal of infectious diseases. During the coronavirus disease 2019 (COVID-19) pandemic, many countries implemented border measures, including restrictions on air travel, to reduce the importation risk of COVID-19. In the context of inbound air travel to Canada, this study aimed to: 1) characterize travel trends before and during the pandemic, and 2) statistically assess the association between travel volumes and travel restrictions during the pandemic. Methods: Monthly commercial air travel volume data from March 2017 to February 2023 were obtained from the International Air Transport Association (IATA). National and airport-level travel trends to Canada were characterized by inbound travel volumes, the number of countries contributing travellers and the ranking of the top ten countries contributing travellers across the study period, by six year-length subperiod groupings (three pre-pandemic and three pandemic). Using seasonal autoregressive integrated moving average (SARIMA) models, interrupted time series (ITS) analyses assessed the association between major travel restrictions and travel volumes by including variables to represent changes to the level and slope of the time series. Results: The pre-pandemic inbound travel volume increased by 3% to 7% between consecutive subperiods, with three seasonal peaks (July-August, December-January, March). At the onset of the pandemic, travel volume decreased by 90%, with the number of contributing countries declining from approximately 200 to 140, followed by a slow recovery in volume and seasonality. A disruption in the ranking of countries that contributed travellers was also noticeable during the pandemic. Results from the ITS analysis aligned with the timing of travel restrictions as follows: implementation in March 2020 coincided with a sharp reduction in volumes, while the easing of major restrictions, starting with the authorization of fully vaccinated travellers from the United States to enter Canada in August 2021, coincided with an increase in the slope of travel volumes. Descriptive and statistical results suggest a near-return of pre-pandemic travel patterns by the end of the study period. Conclusion: Study results suggest resilience in commercial air travel into Canada. Although the COVID-19 pandemic led to a disruption in travel trends, easing of travel restrictions appeared to enable pre-pandemic trends to re-emerge. Understanding trends in air travel volumes, as demonstrated here, can provide information that supports preparedness and response regarding importation risk of infectious pathogens.

One Health communication channels: a qualitative case study of swine influenza in Canada in 2020.

BACKGROUND: With increased attention to the importance of integrating the One Health approach into zoonotic disease surveillance and response, a greater understanding of the mechanisms to support effective communication and information sharing across animal and human health sectors is needed. The objectives of this qualitative case study were to describe the communication channels used between human and animal health stakeholders and to identify the elements that have enabled the integration of the One Health approach. METHODS: We combined documentary research with interviews with fifteen stakeholders to map the communication channels used in human and swine influenza surveillance in Alberta, Canada, as well as in the response to a human case of H1N2v in 2020. A thematic analysis of the interviews was also used to identify the barriers and facilitators to communication among stakeholders from the animal and human health sectors. RESULTS: When a human case of swine influenza emerged, the response led by the provincial Chief Medical Officer of Health involved players at various levels of government and in the human and animal health sectors. The collaboration of public and animal health laboratories and of the swine sector, in addition to the information available through the surveillance systems in place, was swift and effective. Elements identified as enabling smooth communication between the human and animal health systems included preexisting relationships between the various stakeholders, a relationship of trust between them (e.g., the swine sector and their perception of government structures), the presence of stakeholders acting as permanent liaisons between the ministries of health and agriculture, and stakeholders' understanding of the importance of the One Health approach. CONCLUSIONS: Information flows through formal and informal channels and both structural and relational features that can support rapid and effective communication in infectious disease surveillance and outbreak response.

A methodology for estimating SARS-CoV-2 importation risk by air travel into Canada between July and November 2021.

BACKGROUND: Estimating rates of disease importation by travellers is a key activity to assess both the risk to a country from an infectious disease emerging elsewhere in the world and the effectiveness of border measures. We describe a model used to estimate the number of travellers infected with SARS-CoV-2 into Canadian airports in 2021, and assess the impact of pre-departure testing requirements on importation risk. METHODS: A mathematical model estimated the number of essential and non-essential air travellers infected with SARS-CoV-2, with the latter requiring a negative pre-departure test result. The number of travellers arriving infected (i.e. imported cases) depended on air travel volumes, SARS-CoV-2 exposure risk in the departure country, prior infection or vaccine acquired immunity, and, for non-essential travellers, screening from pre-departure molecular testing. Importation risk was estimated weekly from July to November 2021 as the number of imported cases and percent positivity (PP; i.e. imported cases normalised by travel volume). The impact of pre-departure testing was assessed by comparing three scenarios: baseline (pre-departure testing of all non-essential travellers; most probable importation risk given the pre-departure testing requirements), counterfactual scenario 1 (no pre-departure testing of fully vaccinated non-essential travellers), and counterfactual scenario 2 (no pre-departure testing of non-essential travellers). RESULTS: In the baseline scenario, weekly imported cases and PP varied over time, ranging from 145 to 539 cases and 0.15 to 0.28%, respectively. Most cases arrived from the USA, Mexico, the United Kingdom, and France. While modelling suggested that essential travellers had a higher weekly PP (0.37 - 0.65%) than non-essential travellers (0.12 - 0.24%), they contributed fewer weekly cases (62 - 154) than non-essential travellers (84 - 398 per week) given their lower travel volume. Pre-departure testing was estimated to reduce imported cases by one third (counterfactual scenario 1) to one half (counterfactual scenario 2). CONCLUSIONS: The model results highlighted the weekly variation in importation by traveller group (e.g., reason for travel and country of departure) and enabled a framework for measuring the impact of pre-departure testing requirements. Quantifying the contributors of importation risk through mathematical simulation can support the design of appropriate public health policy on border measures.

Forecasting seasonal influenza activity in Canada-Comparing seasonal Auto-Regressive integrated moving average and artificial neural network approaches for public health preparedness.

INTRODUCTION: Public health preparedness is based on timely and accurate information. Time series forecasting using disease surveillance data is an important aspect of preparedness. This study compared two approaches of time series forecasting: seasonal auto-regressive integrated moving average (SARIMA) modelling and the artificial neural network (ANN) algorithm. The goal was to model weekly seasonal influenza activity in Canada using SARIMA and compares its predictive accuracy, based on root mean square prediction error (RMSE) and mean absolute prediction error (MAE), to that of an ANN. METHODS: An initial SARIMA model was fit using automated model selection by minimizing the Akaike information criterion (AIC). Further inspection of the autocorrelation function and partial autocorrelation function led to 'manual' model improvements. ANNs were trained iteratively, using an automated process to minimize the RMSE and MAE. RESULTS: A total of 378, 462 cases of influenza was reported in Canada from the 2010-2011 influenza season to the end of the 2019-2020 influenza season, with an average yearly incidence risk of 20.02 per 100,000 population. Automated SARIMA modelling was the better method in terms of forecasting accuracy (per RMSE and MAE). However, the ANN correctly predicted the peak week of disease incidence while the other models did not. CONCLUSION: Both the ANN and SARIMA models have shown to be capable tools in forecasting seasonal influenza activity in Canada. It was shown that applying both in tandem is beneficial, SARIMA better forecasted overall incidence while ANN correctly predicted the peak week.

An ecological approach to predict areas with established populations of Ixodes scapularis in Quebec, Canada.

Public health management of Lyme disease (LD) is a dynamic challenge in Canada. Climate warming is driving the northward expansion of suitable habitat for the tick vector, Ixodes scapularis. Information about tick population establishment is used to inform the risk of LD but is challenged by sampling biases from surveillance data. Misclassifying areas as having no established tick population underestimates the LD risk classification. We used a logistic regression model at the municipal level to predict the probability of I. scapularis population establishment based on passive tick surveillance data during the period of 2010-2017 in southern Quebec. We tested for the effect of abiotic and biotic factors hypothesized to influence tick biology and ecology. Additional variables controlled for sampling biases in the passive surveillance data. In our final selected model, tick population establishment was positively associated with annual cumulative degree-days > 0°C, precipitation and deer density, and negatively associated with coniferous and mixed forest types. Sampling biases from passive tick surveillance were controlled for using municipal population size and public health instructions on tick submissions. The model performed well as indicated by an area under the curve (AUC) of 0.92, sensitivity of 86% and specificity of 81%. Our model enables prediction of I. scapularis population establishment in areas which lack data from passive tick surveillance and may improve the sensitivity of LD risk categorization in these areas. A more sensitive system of LD risk classification is important for increasing awareness and use of protective measures employed against ticks, and decreasing the morbidity associated with LD.

Effectiveness of non-pharmaceutical interventions to reduce SARS-CoV-2 transmission in Canada and their association with COVID-19 hospitalization rates.

Background: Non-pharmaceutical interventions (NPIs) aim to reduce the incidence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections mostly by limiting contacts between people where virus transmission can occur. However, NPIs limit social interactions and have negative impacts on economic, physical, mental and social well-being. It is, therefore, important to assess the impact of NPIs on reducing the number of coronavirus disease 2019 (COVID-19) cases and hospitalizations to justify their use. Methods: Dynamic regression models accounting for autocorrelation in time series data were used with data from six Canadian provinces (British Columbia, Alberta, Saskatchewan, Manitoba, Ontario, Québec) to assess 1) the effect of NPIs (measured using a stringency index) on SARS-CoV-2 transmission (measured by the effective reproduction number), and 2) the effect of the number of hospitalized COVID-19 patients on the stringency index. Results: Increasing stringency index was associated with a statistically significant decrease in the transmission of SARS-CoV-2 in Alberta, Saskatchewan, Manitoba, Ontario and Québec. The effect of stringency on transmission was time-lagged in all of these provinces except for Ontario. In all provinces except for Saskatchewan, increasing hospitalization rates were associated with a statistically significant increase in the stringency index. The effect of hospitalization on stringency was time-lagged. Conclusion: These results suggest that NPIs have been effective in Canadian provinces, and that their implementation has been, in part, a response to increasing hospitalization rates of COVID-19 patients.

Population surveillance approach to detect and respond to new clusters of COVID-19.

BACKGROUND: To maintain control of the coronavirus disease 2019 (COVID-19) epidemic as lockdowns are lifted, it will be crucial to enhance alternative public health measures. For surveillance, it will be necessary to detect a high proportion of any new cases quickly so that they can be isolated, and people who have been exposed to them traced and quarantined. Here we introduce a mathematical approach that can be used to determine how many samples need to be collected per unit area and unit time to detect new clusters of COVID-19 cases at a stage early enough to control an outbreak. METHODS: We present a sample size determination method that uses a relative weighted approach. Given the contribution of COVID-19 test results from sub-populations to detect the disease at a threshold prevalence level to control the outbreak to 1) determine if the expected number of weekly samples provided from current healthcare-based surveillance for respiratory virus infections may provide a sample size that is already adequate to detect new clusters of COVID-19 and, if not, 2) to determine how many additional weekly samples were needed from volunteer sampling. RESULTS: In a demonstration of our method at the weekly and Canadian provincial and territorial (P/T) levels, we found that only the more populous P/T have sufficient testing numbers from healthcare visits for respiratory illness to detect COVID-19 at our target prevalence level-assumed to be high enough to identify and control new clusters. Furthermore, detection of COVID-19 is most efficient (fewer samples required) when surveillance focuses on healthcare symptomatic testing demand. In the volunteer populations: the higher the contact rates; the higher the expected prevalence level; and the fewer the samples were needed to detect COVID-19 at a predetermined threshold level. CONCLUSION: This study introduces a targeted surveillance strategy, combining both passive and active surveillance samples, to determine how many samples to collect per unit area and unit time to detect new clusters of COVID-19 cases. The goal of this strategy is to allow for early enough detection to control an outbreak.

Ecology of Arctic rabies: 60 years of disease surveillance in the warming climate of northern Canada.

Rabies occurs throughout the Arctic, representing an ongoing public health concern for residents of northern communities. The Arctic fox (Vulpes lagopus) is the main reservoir of the Arctic rabies virus variant, yet little is known about the epidemiology of Arctic rabies, such as the ecological mechanisms driving where and when epizootics in fox populations occur. In this study, we provide the first portrait of the spatio-temporal spread of rabies across northern Canada. We also explore the impact of seasonal and multiannual dynamics in Arctic fox populations and climatic factors on rabies transmission dynamics. We analysed data on rabies cases collected through passive surveillance systems in the Yukon, Northwest Territories, Nunavut, Nunavik and Labrador from 1953 to 2014. In addition, we analysed a large and unique database of trapped foxes tested for rabies in the Northwest Territories and Nunavut from 1974 to 1984 as part of active surveillance studies. Rabies cases occurred in all Arctic regions of Canada and were relatively synchronous among foxes and dogs (Canis familiaris). This study highlights the spread of Arctic rabies virus variant across northern Canada, with contrasting rabies dynamics between different yet connected areas. Population fluctuations of Arctic fox populations could drive rabies transmission dynamics in a complex way across northern Canada. Furthermore, this study suggests different impacts of climate and sea ice cover on the onset of rabies epizootics in northern Canada. These results lay the groundwork for the development of epidemiological models to better predict the spatio-temporal dynamics of rabies occurrence in both wild and domestic carnivores, leading to better estimates of human exposure and transmission risk.

Modeling Mongoose Rabies in the Caribbean: A Model-Guided Fieldwork Approach to Identify Research Priorities.

We applied the model-guided fieldwork framework to the Caribbean mongoose rabies system by parametrizing a spatially-explicit, individual-based model, and by performing an uncertainty analysis designed to identify parameters for which additional empirical data are most needed. Our analysis revealed important variation in output variables characterizing rabies dynamics, namely rabies persistence, exposure level, spatiotemporal distribution, and prevalence. Among epidemiological parameters, rabies transmission rate was the most influential, followed by rabies mortality and location, and size of the initial infection. The most influential landscape parameters included habitat-specific carrying capacities, landscape heterogeneity, and the level of resistance to dispersal associated with topography. Movement variables, including juvenile dispersal, adult fine-scale movement distances, and home range size, as well as life history traits such as age of independence, birth seasonality, and age- and sex-specific mortality were other important drivers of rabies dynamics. We discuss results in the context of mongoose ecology and its influence on disease transmission dynamics. Finally, we suggest empirical approaches and study design specificities that would provide optimal contributing data addressing the knowledge gaps identified by our approach, and would increase our potential to use epidemiological models to guide mongoose rabies control and management in the Caribbean.

Implementing an Action Over Inertia Group Program in Community Residential Rehabilitation Services: Group Participant and Facilitator Perspectives.

Introduction: A time-use focused intervention, Action Over Inertia (AOI) designed to address restricted activity patterns and support recovery, was adapted for use in Australian community residential mental health services. Method: Qualitative case study research explored the use of AOI groups across three Community Care Units from the perspectives of group participants with enduring mental illness and group facilitators. Fifteen interviews were conducted: five group participants were interviewed twice 4 weeks apart, and five group facilitators on completion of the group intervention. Interview data were analyzed thematically using constant comparative methods. Findings: Two overarching themes, "Making Change" and "Facilitating Change" were identified. Efforts to make change in their lives were supported by participants recognising the value of personally meaningful activities for well-being and of activity experiences that fostered hope and recovery, whereas a sense of "stuckness," time for activities and life events could disrupt "getting me going." For the facilitators, facilitating change involved recognizing inertia as a challenge; getting people going; and looking at how AOI intervention works to impact inertia. Conclusion: AOI in a group format supports participants to identify barriers to more active living; to appreciate how time-use and well-being interrelate; and to reframe and take steps to overcome inertia. Further research should evaluate AOI groups as a means of providing individualized support for activity re-engagement as part of recovery oriented mental health rehabilitation.

Application of natural language processing algorithms for extracting information from news articles in event-based surveillance.

The focus of this article is the application of natural language processing (NLP) for information extraction in event-based surveillance (EBS) systems. We describe common information extraction applications from open-source news articles and media sources in EBS systems, methods, value in public health, challenges and emerging developments.

Machine learning and dengue forecasting: Comparing random forests and artificial neural networks for predicting dengue burden at national and sub-national scales in Colombia.

The robust estimate and forecast capability of random forests (RF) has been widely recognized, however this ensemble machine learning method has not been widely used in mosquito-borne disease forecasting. In this study, two sets of RF models were developed at the national (pooled department-level data) and department level in Colombia to predict weekly dengue cases for 12-weeks ahead. A pooled national model based on artificial neural networks (ANN) was also developed and used as a comparator to the RF models. The various predictors included historic dengue cases, satellite-derived estimates for vegetation, precipitation, and air temperature, as well as population counts, income inequality, and education. Our RF model trained on the pooled national data was more accurate for department-specific weekly dengue cases estimation compared to a local model trained only on the department's data. Additionally, the forecast errors of the national RF model were smaller to those of the national pooled ANN model and were increased with the forecast horizon increasing from one-week-ahead (mean absolute error, MAE: 9.32) to 12-weeks ahead (MAE: 24.56). There was considerable variation in the relative importance of predictors dependent on forecast horizon. The environmental and meteorological predictors were relatively important for short-term dengue forecast horizons while socio-demographic predictors were relevant for longer-term forecast horizons. This study demonstrates the potential of RF in dengue forecasting with a feasible approach of using a national pooled model to forecast at finer spatial scales. Furthermore, including sociodemographic predictors is likely to be helpful in capturing longer-term dengue trends.

A call for an ethical framework when using social media data for artificial intelligence applications in public health research.

Advancements in artificial intelligence (AI), more precisely the subfield of machine learning, and their applications to open-source internet data, such as social media, are growing faster than the management of ethical issues for use in society. An ethical framework helps scientists and policy makers consider ethics in their fields of practice, legitimize their work and protect members of the data-generating public. A central question for advancing the ethical framework is whether or not Tweets, Facebook posts and other open-source social media data generated by the public represent a human or not. The objective of this paper is to highlight ethical issues that the public health sector will be or is already confronting when using social media data in practice. The issues include informed consent, privacy, anonymization and balancing these issues with the benefits of using social media data for the common good. Current ethical frameworks need to provide guidance for addressing issues arising from the use of social media data in the public health sector. Discussions in this area should occur while the application of open-source data is still relatively new, and they should also keep pace as other problems arise from ongoing technological change.