A stacked ensemble method for forecasting influenza-like illness visit volumes at emergency departments.
Ontology highlight
ABSTRACT:
SUBMITTER: Novaes de Amorim A
PROVIDER: S-EPMC7984626 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Health planners use forecasts of key metrics associated with influenza-like illness (ILI); near-term weekly incidence, week of season onset, week of peak, and intensity of peak. Here, we describe our participation in a weekly prospective ILI forecasting challenge for the United States for the 2016-17 season and subsequent evaluation of our performance. We implemented a metapopulation model framework with 32 model variants. Variants differed from each other in their assumptions about: the force-of-infection (FOI); use of uninformative priors; the use of discounted historical data for not-yet-observed time points; and the treatment of regions as either independent or coupled. Individual model variants were chosen subjectively as the basis for our weekly forecasts; however, a subset of coupled models were only available part way through the season. Most frequently, during the 2016-17 season, we chose; FOI variants with both school vacations and humidity terms; uninformative priors; the inclusion of discounted historical data for not-yet-observed time points; and coupled regions (when available). Our near-term weekly forecasts substantially over-estimated incidence early in the season when coupled models were not available. However, our forecast accuracy improved in absolute terms and relative to other teams once coupled solutions were available. In retrospective analysis, we found that the 2016-17 season was not typical: on average, coupled models performed better when fit without historically augmented data. Also, we tested a simple ensemble model for the 2016-17 season and found that it underperformed our subjective choice for all forecast targets. In this study, we were able to improve accuracy during a prospective forecasting exercise by coupling dynamics between regions. Although reduction of forecast subjectivity should be a long-term goal, some degree of human intervention is likely to improve forecast accuracy in the medium-term in parallel with the systematic consideration of more sophisticated ensemble approaches.
| S-EPMC6557527 | biostudies-literature
Project description:A variety of mechanistic and statistical methods to forecast seasonal influenza have been proposed and are in use; however, the effects of various data issues and design choices (statistical versus mechanistic methods, for example) on the accuracy of these approaches have not been thoroughly assessed. Here, we compare the accuracy of three forecasting approaches-a mechanistic method, a weighted average of two statistical methods and a super-ensemble of eight statistical and mechanistic models-in predicting seven outbreak characteristics of seasonal influenza during the 2016-2017 season at the national and 10 regional levels in the USA. For each of these approaches, we report the effects of real time under- and over-reporting in surveillance systems, use of non-surveillance proxies of influenza activity and manual override of model predictions on forecast quality. Our results suggest that a meta-ensemble of statistical and mechanistic methods has better overall accuracy than the individual methods. Supplementing surveillance data with proxy estimates generally improves the quality of forecasts and transient reporting errors degrade the performance of all three approaches considerably. The improvement in quality from ad hoc and post-forecast changes suggests that domain experts continue to possess information that is not being sufficiently captured by current forecasting approaches.
| S-EPMC6073642 | biostudies-literature
Project description:Although acute respiratory infections are a leading cause of mortality in sub-Saharan Africa, surveillance of diseases such as influenza is mostly neglected. Evaluating the usefulness of influenza-like illness (ILI) surveillance systems and developing approaches for forecasting future trends is important for pandemic preparedness. We applied and compared a range of robust statistical and machine learning models including random forest (RF) regression, support vector machines (SVM) regression, multivariable linear regression and ARIMA models to forecast 2012 to 2018 trends of reported ILI cases in Cameroon, using Google searches for influenza symptoms, treatments, natural or traditional remedies as well as, infectious diseases with a high burden (i.e., AIDS, malaria, tuberculosis). The R2 and RMSE (Root Mean Squared Error) were statistically similar across most of the methods, however, RF and SVM had the highest average R2 (0.78 and 0.88, respectively) for predicting ILI per 100,000 persons at the country level. This study demonstrates the need for developing contextualized approaches when using digital data for disease surveillance and the usefulness of search data for monitoring ILI in sub-Saharan African countries.
| S-EPMC7991669 | biostudies-literature
Project description:BackgroundPrevious studies of differences in mental health care associated with children's sociodemographic status have focused on access to community care. We examined differences associated with visits to the emergency department.MethodsWe conducted a 6-year population-based cohort analysis using administrative databases of visits (n = 30,656) by children aged less than 18 years (n = 20,956) in Alberta. We measured differences in the number of visits by socioeconomic and First Nations status using directly standardized rates. We examined time to return to the emergency department using a Cox regression model, and we evaluated time to follow-up with a physician by physician type using a competing risks model.ResultsFirst Nations children aged 15-17 years had the highest rate of visits for girls (7047 per 100,000 children) and boys (5787 per 100,000 children); children in the same age group from families not receiving government subsidy had the lowest rates (girls: 2155 per 100,000 children; boys: 1323 per 100,000 children). First Nations children (hazard ratio [HR] 1.64; 95% confidence interval [CI] 1.30-2.05), and children from families receiving government subsidies (HR 1.60, 95% CI 1.30-1.98) had a higher risk of return to an emergency department for mental health care than other children. The longest median time to follow-up with a physician was among First Nations children (79 d; 95% CI 60-91 d); this status predicted longer time to a psychiatrist (HR 0.47, 95% CI 0.32-0.70). Age, sex, diagnosis and clinical acuity also explained post-crisis use of health care.InterpretationMore visits to the emergency department for mental health crises were made by First Nations children and children from families receiving a subsidy. Sociodemographics predicted risk of return to the emergency department and follow-up care with a physician.
| S-EPMC3447077 | biostudies-literature
Project description:In the primary care setting, there are racial and ethnic differences in antibiotic prescribing for acute respiratory tract infections (ARTIs). Viral ARTIs are commonly diagnosed in the pediatric emergency department (PED), in which racial and ethnic differences in antibiotic prescribing have not been previously reported. We sought to investigate whether patient race and ethnicity was associated with differences in antibiotic prescribing for viral ARTIs in the PED. This is a retrospective cohort study of encounters at 7 PEDs in 2013, in which we used electronic health data from the Pediatric Emergency Care Applied Research Network Registry. Multivariable logistic regression was used to examine the association between patient race and ethnicity and antibiotics administered or prescribed among children discharged from the hospital with viral ARTI. Children with bacterial codiagnoses, chronic disease, or who were immunocompromised were excluded. Covariates included age, sex, insurance, triage level, provider type, emergency department type, and emergency department site. Of 39 445 PED encounters for viral ARTIs that met inclusion criteria, 2.6% (95% confidence interval [CI] 2.4%-2.8%) received antibiotics, including 4.3% of non-Hispanic (NH) white, 1.9% of NH black, 2.6% of Hispanic, and 2.9% of other NH children. In multivariable analyses, NH black (adjusted odds ratio [aOR] 0.44; CI 0.36-0.53), Hispanic (aOR 0.65; CI 0.53-0.81), and other NH (aOR 0.68; CI 0.52-0.87) children remained less likely to receive antibiotics for viral ARTIs. Compared with NH white children, NH black and Hispanic children were less likely to receive antibiotics for viral ARTIs in the PED. Future research should seek to understand why racial and ethnic differences in overprescribing exist, including parental expectations, provider perceptions of parental expectations, and implicit provider bias.
| S-EPMC5613999 | biostudies-literature
Project description:The triage process in emergency departments (EDs) relies on the subjective assessment of medical practitioners, making it unreliable in certain aspects. There is a need for a more accurate and objective algorithm to determine the urgency of patients. This paper explores the application of advanced data-synthesis algorithms, machine learning (ML) algorithms, and ensemble models to predict patient mortality. Patients predicted to be at risk of mortality are in a highly critical condition, signifying an urgent need for immediate medical intervention. This paper aims to determine the most effective method for predicting mortality by enhancing the F1 score while maintaining high area under the receiver operating characteristic curve (AUC) score. This study used a dataset of 7325 patients who visited the Yonsei Severance Hospital's ED, located in Seoul, South Korea. The patients were divided into two groups: patients who deceased in the ED and patients who didn't. Various data-synthesis techniques, such as SMOTE, ADASYN, CTGAN, TVAE, CopulaGAN, and Gaussian Copula, were deployed to generate synthetic patient data. Twenty two ML models were then utilized, including tree-based algorithms like Decision tree, AdaBoost, LightGBM, CatBoost, XGBoost, NGBoost, TabNet, which are deep neural network algorithms, and statistical algorithms such as Support Vector Machine, Logistic Regression, Random Forest, k-nearest neighbors, and Gaussian Naive Bayes, as well as Ensemble Models which use the results from the ML models. Based on 21 patient information features used in the pandemic influenza triage algorithm (PITA), the models explained previously were applied to aim for the prediction of patient mortality. In evaluating ML algorithms using an imbalanced medical dataset, conventional metrics like accuracy scores or AUC can be misleading. This paper emphasizes the importance of using the F1 score as the primary performance measure, focusing on recall and specificity in detecting patient mortality. The highest-ranked model for predicting mortality utilized the Gaussian Copula data-synthesis technique and the CatBoost classifier, achieving an AUC of 0.9731 and an F1 score of 0.7059. These findings highlight the effectiveness of machine learning algorithms and data-synthesis techniques in improving the prediction performance of mortality in EDs.
| S-EPMC10497596 | biostudies-literature
Project description:BACKGROUND:Maps of influenza activity are important tools to monitor influenza epidemics and inform policymakers. In France, the availability of a high-quality data set from the Oscour® surveillance network, covering 92% of hospital emergency department (ED) visits, offers new opportunities for disease mapping. Traditional geostatistical mapping methods such as Kriging ignore underlying population sizes, are not suited to non-Gaussian data and do not account for uncertainty in parameter estimates. OBJECTIVE:Our objective was to create reliable weekly interpolated maps of influenza activity in the ED setting, to inform Santé publique France (the French national public health agency) and local healthcare authorities. METHODS:We used Oscour® data of ED visits covering the 2016-2017 influenza season. We developed a Bayesian model-based geostatistical approach, a class of generalized linear mixed models, with a multivariate normal random field as a spatially autocorrelated random effect. Using R-INLA, we developed an algorithm to create maps of the proportion of influenza-coded cases among all coded visits. We compared our results with maps obtained by Kriging. RESULTS:Over the study period, 45 565 (0.82%) visits were coded as influenza cases. Maps resulting from the model are presented for each week, displaying the posterior mean of the influenza proportion and its associated uncertainty. Our model performed better than Kriging. CONCLUSIONS:Our model allows producing smoothed maps where the random noise has been properly removed to reveal the spatial risk surface. The algorithm was incorporated into the national surveillance system to produce maps in real time and could be applied to other diseases.
| S-EPMC6185885 | biostudies-other
Project description:The aim of this study was to determine the age-period-cohort (APC) effects on the rate of infection-related emergency department (ED) visits from septicemia for predicting the same in recent periods.In our study, we investigated the longitudinal trends in septicemia-related visit rates. Using an APC model to decompose the septicemia visit rates into the effects of age, time period, and cohort, and examine whether their effects varied by sex.The septicemia ED visit rate was classified as the International Classification of Disease Code 038 by primary and secondary diagnosis between 1998 and 2012.In both males and females, the visit rate of septicemia showed an increase from 2003 through 2012. An increase in septicemia visit rate after 2003 was observed in all age groups. An APC model indicated a reversal increasing period effect, which increased prominently from 2003 to 2012 in both males and females. The age effect showed an increasing trend. The cohort effect tended to show a slight oscillation from 1913 to 1988. With reference to the prediction of the logarithms of the age-specific 5-year visit rates, we observed that the younger cohorts exhibited a slightly increasing trend, as compared to the older cohorts.The period effect can explain the increase in septicemia visit rates, suggesting the role of screening for septicemia. Furthermore, it is well known that aging is a relevant risk variable for infectious diseases. The present study concludes that the aged population exhibited a strong increasing future trend for septicemia-related ED visit rates.
| S-EPMC5268041 | biostudies-literature
Project description:Influenza-like illness (ILI) is a commonly measured syndromic signal representative of a range of acute respiratory infections. Reliable forecasts of ILI can support better preparation for patient surges in healthcare systems. Although ILI is an amalgamation of multiple pathogens with variable seasonal phasing and attack rates, most existing process-based forecasting systems treat ILI as a single infectious agent. Here, using ILI records and virologic surveillance data, we show that ILI signal can be disaggregated into distinct viral components. We generate separate predictions for six contributing pathogens (influenza A/H1, A/H3, B, respiratory syncytial virus, and human parainfluenza virus types 1-2 and 3), and develop a method to forecast ILI by aggregating these predictions. The relative contribution of each pathogen to the total ILI signal is estimated using a Markov Chain Monte Carlo (MCMC) method upon forecast aggregation. We find highly variable overall contributions from influenza type A viruses across seasons, but relatively stable contributions for the other pathogens. Using historical data from 1997 to 2014 at US national and regional levels, the proposed forecasting system generates improved predictions of both seasonal and near-term targets relative to a baseline method that simulates ILI as a single pathogen. The hierarchical forecasting system can generate predictions for each viral component, as well as infer and predict their contributions to ILI, which may additionally help physicians determine the etiological causes of ILI in clinical settings.
| S-EPMC7608986 | biostudies-literature
Project description:Headache is the most frequent neurological symptom in childhood and the main reason for admission to pediatric emergency departments. The aim of this consensus document is to define a shared clinical pathway between primary care pediatricians (PCP) and hospitals for the management of children presenting with headache. For the purposes of the study, a group of hospital pediatricians and a group of PCP from the Emilia Romagna's health districts were selected to achieve consensus using the RAND/UCLA appropriateness method. Thirty-nine clinical scenarios were developed: for each scenario, participants were asked to rank the appropriateness of each option from 1 to 9. Agreement was reached if ≥75% of participants ranked within the same range of appropriateness. The answers, results, and discussion helped to define the appropriateness of procedures with a low level of evidence regarding different steps of the diagnostic-therapeutic process: primary care evaluation, emergency department evaluation, hospital admission, acute therapy, prophylaxis, and follow-up. The RAND proved to be a valid method to value appropriateness of procedures and define a diagnostic-therapeutic pathway suitable to the local reality in the management of pediatric headache. From our results, some useful recommendations were developed for optimizing the healthcare professionals' network among primary care services and hospitals.
| S-EPMC8877535 | biostudies-literature