Dataset Information

Using search engine big data for predicting new HIV diagnoses.

ABSTRACT:

Background

A large and growing body of "big data" is generated by internet search engines, such as Google. Because people often search for information about public health and medical issues, researchers may be able to use search engine data to monitor and predict public health problems, such as HIV. We sought to assess the feasibility of using Google search data to analyze and predict new HIV diagnoses cases in the United States.

Methods and findings

From 2007 to 2014, we collected search volume data on HIV-related Google search keywords across the United States. State-level new HIV diagnoses data were collected from the Centers for Disease Control and Prevention (CDC) and AIDSVu.org. We developed a negative binomial model to predict HIV cases using a subset of significant predictor keywords identified by LASSO. The Google search data were combined with state-level HIV case reports provided by the CDC. We use historical data to train the model and predict new HIV diagnoses from 2011 to 2014, with an average R2 value of 0.99 between predicted versus actual cases, and average root-mean-square error (RMSE) of 108.75.

Conclusions

Results indicate that Google Trends is a feasible tool to predict new cases of HIV at the state level. We discuss the implications of integrating visualization maps and tools based on these models into public health and HIV monitoring and surveillance.

SUBMITTER: Young SD

PROVIDER: S-EPMC6042696 | biostudies-literature | 2018

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Using search engine big data for predicting new HIV diagnoses.

Young Sean D SD Zhang Qingpeng Q

PloS one 20180712 7

<h4>Background</h4>A large and growing body of "big data" is generated by internet search engines, such as Google. Because people often search for information about public health and medical issues, researchers may be able to use search engine data to monitor and predict public health problems, such as HIV. We sought to assess the feasibility of using Google search data to analyze and predict new HIV diagnoses cases in the United States.<h4>Methods and findings</h4>From 2007 to 2014, we collecte ...[more]

PMID: 30001360

Similar Datasets

Project description:BackgroundGiven the ongoing COVID-19 pandemic situation, accurate predictions could greatly help in the health resource management for future waves. However, as a new entity, COVID-19's disease dynamics seemed difficult to predict. External factors, such as internet search data, need to be included in the models to increase their accuracy. However, it remains unclear whether incorporating online search volumes into models leads to better predictive performances for long-term prediction.ObjectiveThe aim of this study was to analyze whether search engine query data are important variables that should be included in the models predicting new daily COVID-19 cases and deaths in short- and long-term periods.MethodsWe used country-level case-related data, NAVER search volumes, and mobility data obtained from Google and Apple for the period of January 20, 2020, to July 31, 2021, in South Korea. Data were aggregated into four subsets: 3, 6, 12, and 18 months after the first case was reported. The first 80% of the data in all subsets were used as the training set, and the remaining data served as the test set. Generalized linear models (GLMs) with normal, Poisson, and negative binomial distribution were developed, along with linear regression (LR) models with lasso, adaptive lasso, and elastic net regularization. Root mean square error values were defined as a loss function and were used to assess the performance of the models. All analyses and visualizations were conducted in SAS Studio, which is part of the SAS OnDemand for Academics.ResultsGLMs with different types of distribution functions may have been beneficial in predicting new daily COVID-19 cases and deaths in the early stages of the outbreak. Over longer periods, as the distribution of cases and deaths became more normally distributed, LR models with regularization may have outperformed the GLMs. This study also found that models performed better when predicting new daily deaths compared to new daily cases. In addition, an evaluation of feature effects in the models showed that NAVER search volumes were useful variables in predicting new daily COVID-19 cases, particularly in the first 6 months of the outbreak. Searches related to logistical needs, particularly for "thermometer" and "mask strap," showed higher feature effects in that period. For longer prediction periods, NAVER search volumes were still found to constitute an important variable, although with a lower feature effect. This finding suggests that search term use should be considered to maintain the predictive performance of models.ConclusionsNAVER search volumes were important variables in short- and long-term prediction, with higher feature effects for predicting new daily COVID-19 cases in the first 6 months of the outbreak. Similar results were also found for death predictions.

Project description:ObjectivesHand, foot, and mouth disease (HFMD) has caused a substantial burden in China, especially in Guangdong Province. Based on the enhanced surveillance system, we aimed to explore whether the addition of temperate and search engine query data improves the risk prediction of HFMD.DesignEcological study.Setting and participantsInformation on the confirmed cases of HFMD, climate parameters and search engine query logs was collected. A total of 1.36 million HFMD cases were identified from the surveillance system during 2011-2014. Analyses were conducted at aggregate level and no confidential information was involved.Outcome measuresA seasonal autoregressive integrated moving average (ARIMA) model with external variables (ARIMAX) was used to predict the HFMD incidence from 2011 to 2014, taking into account temperature and search engine query data (Baidu Index, BDI). Statistics of goodness-of-fit and precision of prediction were used to compare models (1) based on surveillance data only, and with the addition of (2) temperature, (3) BDI, and (4) both temperature and BDI.ResultsA high correlation between HFMD incidence and BDI (r=0.794, p<0.001) or temperature (r=0.657, p<0.001) was observed using both time series plot and correlation matrix. A linear effect of BDI (without lag) and non-linear effect of temperature (1 week lag) on HFMD incidence were found in a distributed lag non-linear model. Compared with the model based on surveillance data only, the ARIMAX model including BDI reached the best goodness-of-fit with an Akaike information criterion (AIC) value of -345.332, whereas the model including both BDI and temperature had the most accurate prediction in terms of the mean absolute percentage error (MAPE) of 101.745%.ConclusionsAn ARIMAX model incorporating search engine query data significantly improved the prediction of HFMD. Further studies are warranted to examine whether including search engine query data also improves the prediction of other infectious diseases in other settings.

Dataset Information

Using search engine big data for predicting new HIV diagnoses.

Background

Methods and findings

Conclusions

Publications

Using search engine big data for predicting new HIV diagnoses.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets