Dataset Information

Improved forecasts of influenza-associated hospitalization rates with Google Search Trends.

ABSTRACT: Reliable forecasts of influenza-associated hospitalizations during seasonal outbreaks can help health systems better prepare for patient surges. Within the USA, public health surveillance systems collect and distribute near real-time weekly hospitalization rates, a key observational metric that makes real-time forecast of this outcome possible. In this paper, we describe a method to forecast hospitalization rates using a population level transmission model in combination with a data assimilation technique. Using this method, we generated retrospective forecasts of hospitalization rates for five age groups and the overall population during five seasons in the USA and quantified forecast accuracy for both near-term and seasonal targets. Additionally, we describe methods to correct for under-reporting of hospitalization rates (backcast) and to estimate hospitalization rates from publicly available online search trends data (nowcast). Forecasts based on surveillance rates alone were reasonably accurate in predicting peak hospitalization rates (within ± 25% of the actual peak rate, three weeks before peak). The error in predicting rates one to four weeks ahead, remained constant for the duration of the seasons, even during periods of increased influenza incidence. An improvement in forecast quality across all age groups, seasons and targets was observed when backcasts and nowcasts supplemented surveillance data. These results suggest that the model-inference framework can provide reasonably accurate real-time forecasts of influenza hospitalizations; backcasts and nowcasts offer a way to improve system tolerance to observational errors.

SUBMITTER: Kandula S

PROVIDER: S-EPMC6597779 | biostudies-literature | 2019 Jun

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Improved forecasts of influenza-associated hospitalization rates with Google Search Trends.

Kandula Sasikiran S Pei Sen S Shaman Jeffrey J

Journal of the Royal Society, Interface 20190612 155

Reliable forecasts of influenza-associated hospitalizations during seasonal outbreaks can help health systems better prepare for patient surges. Within the USA, public health surveillance systems collect and distribute near real-time weekly hospitalization rates, a key observational metric that makes real-time forecast of this outcome possible. In this paper, we describe a method to forecast hospitalization rates using a population level transmission model in combination with a data assimilation ...[more]

PMID: 31185818

Similar Datasets

Project description:BackgroundSince the beginning of the new coronavirus pandemic, there has been much information about the disease and the virus has been in the spotlight, shared and commented upon on the Internet. However, much of this information is infodemics and can interfere with the advancement of the disease and that way that populations act. Thus, Brazil is a country that requires attention, as despite the fact that in almost two years of pandemic it has shown a devastating numbers of deaths and number of cases, and generates false, distorted and malicious news about the pandemic. This work intends to understand the attitudes of the Brazilian population using infodemic queries from the Google Trends search tool and social and income variables.MethodsData from infodemic research carried out on Google Trends, between January 1, 2020 and June 30, 2021, with socioeconomic data, such as income and education, were unified in a single database: standardization and exploratory and multivalued techniques based on grouping were used in the study.ResultsIn the analysis of the search trend of infodemic terms, it is clear that the categories of Prevention and Beliefs should stand out in Brazil, where there is a diverse culture. It is followed by the COVID-19 Treatment category, with treatments that were not those recommended by the authorities. Income transfer programs and information on socioeconomic variables did not have much impact on infodemic surveys, but it was observed that states where President Bolsonaro has more supporters had researched more infodemic information.ConclusionsIn a country as geographically large as Brazil, it is important that political authorities go to great lengths to disseminate reliable information and monitor the infodemic in the media and on the internet. It was concluded that the denial of the pandemic and the influence of political leaders influenced the search for infodemic information, contributing to a disorganization in the control of the disease and prevention measures.

Project description:ImportanceAfter abrupt closures of businesses and public gatherings in the US in late spring 2020 due to the COVID-19 pandemic, by mid-May 2020, most states reopened their economies. Owing in part to a lack of earlier data, there was little evidence on whether state reopening policies influenced important pandemic outcomes-COVID-19-related hospitalizations and mortality-to guide future decision-making in the remainder of this and future pandemics.ObjectiveTo investigate changes in COVID-19-related hospitalizations and mortality trends after reopening of US state economies.Design setting and participantsUsing an interrupted time series approach, this cross-sectional study examined trends in per-capita COVID-19-related hospitalizations and deaths before and after state reopenings between April 16 and July 31, 2020. Daily state-level data from the University of Minnesota COVID-19 Hospitalization Tracking Project on COVID-19-related hospitalizations and deaths across 47 states were used in the analysis.ExposuresDates that states reopened their economies.Main outcomes and measuresState-day observations of COVID-19-related hospitalizations and COVID-19-related new deaths per 100 000 people.ResultsThe study included 3686 state-day observations of hospitalizations and 3945 state-day observations of deaths. On the day of reopening, the mean number of hospitalizations per 100 000 people was 17.69 (95% CI, 12.54-22.84) and the mean number of daily new deaths per 100 000 people was 0.395 (95% CI, 0.255-0.536). Both outcomes displayed flat trends before reopening, but they started trending upward thereafter. Relative to the hospitalizations trend in the period before state reopenings, the postperiod trend was higher by 1.607 per 100 000 people (95% CI, 0.203-3.011; P = .03). This estimate implied that nationwide reopenings were associated with 5319 additional people hospitalized for COVID-19 each day. The trend in new deaths after reopening was also positive (0.0376 per 100 000 people; 95% CI, 0.0038-0.0715; P = .03), but the change in mortality trend was not significant (0.0443; 95% CI, -0.0048 to 0.0933; P = .08).Conclusions and relevanceIn this cross-sectional study conducted over a 3.5-month period across 47 US states, data on the association of hospitalizations and mortality with state reopening policies may provide input to state projections of the pandemic as policy makers continue to balance public health protections with sustaining economic activity.

Project description:BackgroundThe objective of this study is to investigate predictive utility of online social media and web search queries, particularly, Google search data, to forecast new cases of influenza-like-illness (ILI) in general outpatient clinics (GOPC) in Hong Kong. To mitigate the impact of sensitivity to self-excitement (i.e., fickle media interest) and other artifacts of online social media data, in our approach we fuse multiple offline and online data sources.MethodsFour individual models: generalized linear model (GLM), least absolute shrinkage and selection operator (LASSO), autoregressive integrated moving average (ARIMA), and deep learning (DL) with Feedforward Neural Networks (FNN) are employed to forecast ILI-GOPC both one week and two weeks in advance. The covariates include Google search queries, meteorological data, and previously recorded offline ILI. To our knowledge, this is the first study that introduces deep learning methodology into surveillance of infectious diseases and investigates its predictive utility. Furthermore, to exploit the strength from each individual forecasting models, we use statistical model fusion, using Bayesian model averaging (BMA), which allows a systematic integration of multiple forecast scenarios. For each model, an adaptive approach is used to capture the recent relationship between ILI and covariates.ResultsDL with FNN appears to deliver the most competitive predictive performance among the four considered individual models. Combing all four models in a comprehensive BMA framework allows to further improve such predictive evaluation metrics as root mean squared error (RMSE) and mean absolute predictive error (MAPE). Nevertheless, DL with FNN remains the preferred method for predicting locations of influenza peaks.ConclusionsThe proposed approach can be viewed a feasible alternative to forecast ILI in Hong Kong or other countries where ILI has no constant seasonal trend and influenza data resources are limited. The proposed methodology is easily tractable and computationally efficient.

Project description:BACKGROUND:Limiting the adverse effects of seasonal influenza outbreaks at state or city level requires close monitoring of localized outbreaks and reliable forecasts of their progression. Whereas forecasting models for influenza or influenza-like illness (ILI) are becoming increasingly available, their applicability to localized outbreaks is limited by the nonavailability of real-time observations of the current outbreak state at local scales. Surveillance data collected by various health departments are widely accepted as the reference standard for estimating the state of outbreaks, and in the absence of surveillance data, nowcast proxies built using Web-based activities such as search engine queries, tweets, and access of health-related webpages can be useful. Nowcast estimates of state and municipal ILI were previously published by Google Flu Trends (GFT); however, validations of these estimates were seldom reported. OBJECTIVE:The aim of this study was to develop and validate models to nowcast ILI at subregional geographic scales. METHODS:We built nowcast models based on autoregressive (autoregressive integrated moving average; ARIMA) and supervised regression methods (Random forests) at the US state level using regional weighted ILI and Web-based search activity derived from Google's Extended Trends application programming interface. We validated the performance of these methods using actual surveillance data for the 50 states across six seasons. We also built state-level nowcast models using state-level estimates of ILI and compared the accuracy of these estimates with the estimates of the regional models extrapolated to the state level and with the nowcast estimates published by GFT. RESULTS:Models built using regional ILI extrapolated to state level had a median correlation of 0.84 (interquartile range: 0.74-0.91) and a median root mean square error (RMSE) of 1.01 (IQR: 0.74-1.50), with noticeable variability across seasons and by state population size. Model forms that hypothesize the availability of timely state-level surveillance data show significantly lower errors of 0.83 (0.55-0.23). Compared with GFT, the latter model forms have lower errors but also lower correlation. CONCLUSIONS:These results suggest that the proposed methods may be an alternative to the discontinued GFT and that further improvements in the quality of subregional nowcasts may require increased access to more finely resolved surveillance data.

Dataset Information

Improved forecasts of influenza-associated hospitalization rates with Google Search Trends.

Publications

Improved forecasts of influenza-associated hospitalization rates with Google Search Trends.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets