Towards Identifying and Reducing the Bias of Disease Information Extracted from Search Engine Data

Early warning and predicting of COVID-19 using zero-inflated negative binomial regression model and negative binomial regression model

BACKGROUND

It is difficult to detect the outbreak of emergency infectious disease based on the exiting surveillance system. Here we investigate the utility of the Baidu Search Index, an indicator of how large of a keyword is in Baidu's search volume, in the early warning and predicting the epidemic trend of COVID-19.

METHODS

The daily number of cases and the Baidu Search Index of 8 keywords (weighted by population) from December 1, 2019 to March 15, 2020 were collected and analyzed with times series and Spearman correlation with different time lag. To predict the daily number of COVID-19 cases using the Baidu Search Index, Zero-inflated negative binomial regression was used in phase 1 and negative binomial regression model was used in phase 2 and phase 3 based on the characteristic of independent variable.

RESULTS

The Baidu Search Index of all keywords in Wuhan was significantly higher than Hubei (excluded Wuhan) and China (excluded Hubei). Before the causative pathogen was identified, the search volume of "Influenza" and "Pneumonia" in Wuhan increased with the number of new onset cases, their correlation coefficient was 0.69 and 0.59, respectively. After the pathogen was public but before COVID-19 was classified as a notifiable disease, the search volume of "SARS", "Pneumonia", "Coronavirus" in all study areas increased with the number of new onset cases with the correlation coefficient was 0.69 ~ 0.89, while "Influenza" changed to negative correlated (rs: -0.56 ~ -0.64). After COVID-19 was closely monitored, the Baidu Search Index of "COVID-19", "Pneumonia", "Coronavirus", "SARS" and "Mask" could predict the epidemic trend with 15 days, 5 days and 6 days lead time, respectively in Wuhan, Hubei (excluded Wuhan) and China (excluded Hubei). The predicted number of cases would increase 1.84 and 4.81 folds, respectively than the actual number of cases in Wuhan and Hubei (excluded Wuhan) from 21 January to 9 February.

CONCLUSION

The Baidu Search Index could be used in the early warning and predicting the epidemic trend of COVID-19, but the search keywords changed in different period. Considering the time lag from onset to diagnosis, especially in the areas with medical resources shortage, internet search data can be a highly effective supplement of the existing surveillance system.

Analysing the impact of comorbid conditions and media coverage on online symptom search data: a novel AI-based approach for COVID-19 tracking

BACKGROUND

Web search data have proven to bea valuable early indicator of COVID-19 outbreaks. However, the influence of co-morbid conditions with similar symptoms and the effect of media coverage on symptom-related searches are often overlooked, leading to potential inaccuracies in COVID-19 simulations.

METHOD

This study introduces a machine learning-based approach to estimate the magnitude of the impact of media coverage and comorbid conditions with similar symptoms on online symptom searches, based on two scenarios with quantile levels 10-90 and 25-75. An incremental batch learning RNN-LSTM model was then developed for the COVID-19 simulation in Australia and New Zealand, allowing the model to dynamically simulate different infection rates and transmissibility of SARS-CoV-2 variants.

RESULT

The COVID-19 infected person-directed symptom searches were found to account for only a small proportion of the total search volume (on average 33.68% in Australia vs. 36.89% in New Zealand) compared to searches influenced by media coverage and comorbid conditions (on average 44.88% in Australia vs. 50.94% in New Zealand). The proposed method, which incorporates estimated symptom component ratios into the RNN-LSTM embedding model, significantly improved COVID-19 simulation performance.

CONCLUSION

Media coverage and comorbid conditions with similar symptoms dominate the total number of online symptom searches, suggesting that direct use of online symptom search data in COVID-19 simulations may overestimate COVID-19 infections. Our approach provides new insights into the accurate estimation of COVID-19 infections using online symptom searches, thereby assisting governments in developing complementary methods for public health surveillance.

Increasing public concern on insomnia during the COVID-19 outbreak in China: An info-demiology study

Background

Since December 2019, an unexplained pneumonia has broken out in Wuhan, Hubei Province, China. In order to prevent the rapid spread of this disease, quarantine or lockdown measures were taken by the Chinese government. These measures turned out to be effective in containing the contagious disease. In spite of that, social distancing measures, together with disease itself, would potentially cause certain health risks among the affected population, such as sleep disorder. We herein conducted this web search analysis so as to examine the temporal and spatial changes of public search volume of the mental health topic of "insomnia" during COVID-19 pandemic in China.

Methods

The data sources included Baidu Index (BDI) to analyze related search terms and the official website of the National Health Commission of the People's Republic of China to collect the daily number of newly confirmed COVID-19 cases. Following a descriptive analysis of the overall search situation, Spearman's correlation analysis was used to analyze the relationship between the daily insomnia-related search values and the daily newly confirmed cases. The means of search volume for insomnia-related terms during the COVID-19 outbreak period (January 23rd, 2020 to April 8th, 2020) were compared with those during 2016-2019 using Student's t test. Finally, by analyzing the overall daily mean of insomnia in various provinces, we further evaluated whether there existed regional differences in searching for insomnia during the COVID-19 outbreak period.

Results

During the COVID-19 outbreak period, the number of insomnia-related searches increased significantly, especially the average daily the BDI for the term "1 min to fall asleep immediately". Spearman's correlation analysis showed that 6 out of the 10 insomnia-related keywords were significantly positively related to the daily newly confirmed cases. Compared with the same period in the past four years, a significantly increased search volume was found in 60.0% (6/10) insomnia-related terms during the COVID-19 outbreak period. We also found that Guangdong province had the highest number of searches for insomnia-related during the pandemic.

Conclusions

The surge in the number of confirmed cases during the COVID-19 pandemic has led to an increase in concern and online searches on this topic of insomnia. Further studies are needed to determine whether the search behavior truly reflect the real-time prevalence profile of relevant mental disorders, and further to establish a risk prediction model to determine the prevalence risk of psychopathological disorders, including insomnia, using insomnia-related BDI and other well-established risk factors.

Application of the Internet Platform in Monitoring Chinese Public Attention to the Outbreak of COVID-19

Objectives

The internet data is an essential tool for reflecting public attention to hot issues. This study aimed to use the Baidu Index (BDI) and Sina Micro Index (SMI) to confirm correlation between COVID-19 case data and Chinese online data (public attention). This could verify the effect of online data on early warning of public health events, which will enable us to respond in a more timely and effective manner.

Methods

Spearman correlation was used to check the consistency of BDI and SMI. Time lag cross-correlation analysis of BDI, SMI and six case-related indicators and multiple linear regression prediction were performed to explore the correlation between public concern and the actual epidemic.

Results

The public's usage trend of the Baidu search engine and Sina Weibo was consistent during the COVID-19 outbreak. BDI, SMI and COVID-19 indicators had significant advance or lag effects, among which SMI and six indicators all had advance effects while BDI only had advance effects with new confirmed cases and new death cases. But compared with the SMI, the BDI was more closely related to the epidemic severity. Notably, the prediction model constructed by BDI and SMI can well fit new confirmed cases and new death cases.

Conclusions

The confirmed associations between the public's attention to the outbreak of COVID and the trend of epidemic outbreaks implied valuable insights into effective mechanisms of crisis response. In response to public health emergencies, people can through the information recommendation functions of social media and search engines (such as Weibo hot search and Baidu homepage recommendation) to raise awareness of available disease prevention and treatment, health services, and policy change.

Social Media- and Internet-Based Disease Surveillance for Public Health

Disease surveillance systems are a cornerstone of public health tracking and prevention. This review addresses the use, promise, perils, and ethics of social media- and Internet-based data collection for public health surveillance. Our review highlights untapped opportunities for integrating digital surveillance in public health and current applications that could be improved through better integration, validation, and clarity on rules surrounding ethical considerations. Promising developments include hybrid systems that couple traditional surveillance data with data from search queries, social media posts, and crowdsourcing. In the future, it will be important to identify opportunities for public and private partnerships, train public health experts in data science, reduce biases related to digital data (gathered from Internet use, wearable devices, etc.), and address privacy. We are on the precipice of an unprecedented opportunity to track, predict, and prevent global disease burdens in the population using digital data.

The application of meteorological data and search index data in improving the prediction of HFMD: A study of two cities in Guangdong Province, China

Hand, foot and mouth disease (HFMD) is a public health issue in China, and its incidence in Guangdong Province is higher than the national average. Previous studies have found climatic factors have an influential role in the transmission of HFMD. Internet search technology has been shown to predict some infectious disease epidemics and is a potential resource in tracking epidemics in countries where the use of Internet search index data is prevalent. This study aims to improve the prediction of HFMD in two Chinese cities, Shantou and Shenzhen in Guangdong Province, applying both meteorological data and Baidu search indices to create a HFMD forecasting model. To this end, the relationship between meteorological factors and HFMD was found to be linear in both cities, while the relationship between search engine data and HFMD was not consistent. The results of our study suggest that using both Internet search and meteorological data can improve the prediction of HFMD incidence. Using comparative analysis of both cities, we posit that improved quality search indices enhance prediction of HFMD.

Characteristics of public concern on haze in China and its relationship with air quality in urban areas

Severe air pollution associated with the rapid urbanization is a pressing issue in China. Moreover, the public awareness of environmental protection in China is awakening, which poses enormous pressure on governments to enforce environmental regulations. The study of environmental problems from the public perspective plays a crucial role in effective environmental governance. The Baidu search engine is the China's largest search engine. The search index of haze based on Baidu search engine reflects the public concern on air quality in China. The aim of this study is to uncover important relationships between public concern and air quality monitoring data based on the case study of haze pollution crisis in China. The results indicate that: (1) the year 2013 is the turning point of the public concern on air quality in China; (2) according to daily data analysis, the search index of haze has increased progressively with increased PM_2.5 concentration with a time lag of 0-4 days and the lag time has a declining tendency from 2013 to 2017; (3) according to annual data analysis, the public concern showed a weak correlation with air quality and they showed an opposite temporal trend. However, when the long-term annual trend was removed, the strong positive correlation emerges between the fluctuation parts of the search index of haze and monitoring data of air quality. This indicates the public is more sensitive to the short-term fluctuation of air quality. The results of this paper provide statistical evidence on the evolution of public concern on air quality from 2013 to 2017. This study will help policy makers to better understand the patterns of the public's perception of environmental problems and consequently improve the government's capability to deal with these challenges.

Using Baidu index to nowcast hand-foot-mouth disease in China: a meta learning approach

BACKGROUND

Hand, foot, and mouth disease (HFMD) has been recognized as one of the leading infectious diseases among children in China, which causes hundreds of annual deaths since 2008. In China, the reports of monthly HFMD cases usually have a delay of 1-2 months due to the time needed for collecting and processing clinical information. This time lag is far from optimal for policymakers making decisions. To alleviate this information gap, this study uses a meta learning framework and combines publicly Internet-based information (Baidu search queries) for real-time estimation of HFMD cases.

METHODS

We incorporate Baidu index into modeling to nowcast the monthly HFMD incidences in Guangxi, Zhejiang, Henan provinces and the whole China. We develop a meta learning framework to select appropriate predictive model based on the statistical and time series meta features. Our proposed approach is assessed for the HFMD cases within the time period from July 2015 to June 2016 using multiple evaluation metrics including root mean squared error (RMSE) and correlation coefficient (Corr).

RESULTS

For the four areas: whole China, Guangxi, Zhejiang, and Henan, our approach is superior to the best competing models, reducing the RMSE by 37, 20, 20, and 30% respectively. Compared with all the alternative predictive methods, our estimates show the strongest correlation with the observations.

CONCLUSIONS

In this study, the proposed meta learning method significantly improves the HFMD prediction accuracy, demonstrating that: (1) the Internet-based information offers the possibility for effective HFMD nowcasts; (2) the meta learning approach is capable of adapting to a wide variety of data, and enables selecting appropriate method for improving the nowcasting accuracy.

Modeling and inference for infectious disease dynamics: a likelihood-based approach

Likelihood-based statistical inference has been considered in most scientific fields involving stochastic modeling. This includes infectious disease dynamics, where scientific understanding can help capture biological processes in so-called mechanistic models and their likelihood functions. However, when the likelihood of such mechanistic models lacks a closed-form expression, computational burdens are substantial. In this context, algorithmic advances have facilitated likelihood maximization, promoting the study of novel data-motivated mechanistic models over the last decade. Reviewing these models is the focus of this paper. In particular, we highlight statistical aspects of these models like overdispersion, which is key in the interface between nonlinear infectious disease modeling and data analysis. We also point out potential directions for further model exploration.

Monitoring hand, foot and mouth disease by combining search engine query data and meteorological factors

Hand, foot and mouth disease (HFMD) has been recognized as a significant public health threat and poses a tremendous challenge to disease control departments. To date, the relationship between meteorological factors and HFMD has been documented, and public interest of disease has been proven to be trackable from the Internet. However, no study has explored the combination of these two factors in the monitoring of HFMD. Therefore, the main aim of this study was to develop an effective monitoring model of HFMD in Guangzhou, China by utilizing historical HFMD cases, Internet-based search engine query data and meteorological factors. To this end, a case study was conducted in Guangzhou, using a network-based generalized additive model (GAM) including all factors related to HFMD. Three other models were also constructed using some of the variables for comparison. The results suggested that the model showed the best estimating ability when considering all of the related factors.