Internet Search Behavior Research Articles

Pandemic-induced fear and stock market returns: Evidence from China

We construct a pandemic-induced fear (PIF) index to measure fear of the COVID-19 pandemic using Internet search volumes of the Chinese local search engine and empirically investigate the impact of fear of the pandemic on Chinese stock market returns. A reduced-bias estimation approach for multivariate regression is employed to address the issue of small-sample bias. We find that the PIF index has a negative and significant impact on cumulative stock market returns. The impact of PIF is persistent, which can be explained by mispricing from investors' excessive pessimism. We further reveal that the PIF index directly predicts stock market returns through noise trading. Investors' Internet search behaviors enhance the fear of the pandemic, and pandemic-induced fear determines future stock market returns, rather than the number of cases and deaths caused by the COVID-19 pandemic.

Impacts of the COVID-19 pandemic on human-nature interactions: Pathways, evidence and implications.

The coronavirus (COVID‐19) pandemic and the global response have dramatically changed people's lifestyles in much of the world. These major changes, as well as the associated changes in impacts on the environment, can alter the dynamics of the direct interactions between humans and nature (hereafter human–nature interactions) far beyond those concerned with animals as sources of novel human coronavirus infections. There may be a variety of consequences for both people and nature.Here, we suggest a conceptual framework for understanding how the COVID‐19 pandemic might affect the dynamics of human–nature interactions. This highlights three different, but not mutually exclusive, pathways: changes in (a) opportunity, (b) capability and (c) motivation.Through this framework, we also suggest that there are several feedback loops by which changes in human–nature interactions induced by the COVID‐19 pandemic can lead to further changes in these interactions such that the impacts of the pandemic could persist over the long term, including after it has ended.The COVID‐19 pandemic, which has had the most tragic consequences, can also be viewed as a ‘global natural experiment’ in human–nature interactions that can provide unprecedented mechanistic insights into the complex processes and dynamics of these interactions and into possible strategies to manage them to best effect. A free Plain Language Summary can be found within the Supporting Information of this article.

Google search trends for itch in Europe: a retrospective longitudinal study.

Itch is a common symptom in the general population. Affected individuals often do not seek medical consultation and rely on Internet searches to obtain information regarding their itch. The aim of this study was to attain insights into common concerns of the general population regarding itch can by analysing itch-related Internet search behaviour. Google AdWords Keyword Planner was used to assess search volumes for itch-related terms in 15 European countries between September 2014 and August 2018. All identified keywords were qualitatively categorized. Itch-related terms were descriptively analysed and are shown as number of searches/100000 inhabitants. The search volume for the keyword 'itch' per 100000 inhabitants was highest in Northern Europe, followed by Eastern, Central and Southern Europe. In 4/15 countries, itch was searched for more often in the autumn/winter months compared to in the spring/summer months. Most itch-related terms were related to dermatological conditions such as inflammatory skin diseases (e.g. psoriasis, atopic dermatitis), allergic or immunologic conditions (e.g. urticaria), and infectious diseases or infestations (e.g. scabies). In terms of body location, genitoanal itch dominated the searches. Symptoms and signs related to itch, possible non-dermatological aetiologies, and treatment options were also among the most searched terms. These analyses provided for the first time insights into the search behaviour patterns related to itch across Europe. People from Northern and Eastern Europe are more likely to seek online information regarding itch. Causes for the itch, especially dermatological conditions, and genitoanal itch are the most important concerns for Internet users. This unconventional and inexpensive method identifies medical needs of people beyond the medical setting, including people who do not seek medical consultation. Accordingly, the data could be used to guide public health interventions and manage respective inhabitants' medical needs.

Changes in internet search behaviors related to smell and taste disorders during the COVID-19 outbreak in Turkey

Objectives: This study aims to assess the change in internet searches terms related to smell and taste disorders (STDs) during the novel coronavirus-2019 (COVID-19) outbreak and to investigate the relationship between the changes and the daily number of COVID-19 cases and deaths reported in Turkey. Patients and Methods: Between March 1st, 2019 and April 30th, 2020, the average number of internet searches of the most common 10 key terms related to STDs during the COVID-19 outbreak was compared to the average number of internet searches of those terms in previous months in Turkey. The daily average number of most common 10 key terms based on Google AdWords (GA) and the relative search volume (RSV) of most searched two terms (“inability to smell” and “inability to taste”) related to STDs based on Google trends (GT) were compared with the change in the daily number of COVID-19 cases and deaths in Turkey. Results: The average number of internet searches of 10 key terms between March and April 2020 increased approximately six times from both the average number of the same terms in the previous 10 months and between March and April 2019 months period (p<0.001). We identified a strong correlation between the daily numerical changes of key terms and the increase in the number of daily COVID-19 cases and deaths (r=0.810 and r=0.783, respectively; p<0.001). There was a strong correlation between the RSV surges of two terms (“inability to smell” and “inability to taste”) and the increase in the daily number of cases and deaths during the outbreak in Turkey (p<0.001). Conclusion: The COVID-19 in Turkey led to an increase in internet searches terms related to STDs. This increase was strongly correlated with the increase in COVID-19 cases and deaths in Turkey.

Prediction of ambulatory visits for itching diseases using Google Trends data

It would provide insight to know if Google queries for “itch” relate to the probability of a patient visit for common itching diseases in the National Ambulatory Medical Care Survey. There was a negative predictive value of visits for vaginitis, vulvitis, and vulvovaginitis [odds ratio 0.983 (95% confidence interval 0.977, 0.989)], psoriasis [0.986 (0.980, 0.992)], and dermatitis [0.983 (0.979, 0.986)] with increased normalized search interest (P<0.0001). There was no relationship between search interest and visits for urticaria [0.997 (0.986, 1.01), P=0.5] or atopic dermatitis [1.02 (0.994, 1.01), P=0.9]. This demonstrates that internet search behavior can affect real world patient visits.

Impact of online mental health screening tools on help-seeking, care receipt, and suicidal ideation and suicidal intent: Evidence from internet search behavior in a large U.S. cohort

IntroductionMost people with psychiatric illnesses do not receive treatment for almost a decade after disorder onset. Online mental health screens reflect one mechanism designed to shorten this lag in help-seeking, yet there has been limited research on the effectiveness of screening tools in naturalistic settings. Material and methodsWe examined a cohort of persons directed to a mental health screening tool via the Bing search engine (n = 126,060). We evaluated the impact of tool content on later searches for mental health self-references, self-diagnosis, care seeking, psychoactive medications, suicidal ideation, and suicidal intent. Website characteristics were evaluated by pairs of independent raters to ascertain screen type and content. These included the presence/absence of a suggestive diagnosis, a message on interpretability, as well as referrals to digital treatments, in-person treatments, and crisis services. ResultsUsing machine learning models, the results suggested that screen content predicted later searches with mental health self-references (AUC = 0·73), mental health self-diagnosis (AUC = 0·69), mental health care seeking (AUC = 0·61), psychoactive medications (AUC = 0·55), suicidal ideation (AUC = 0·58), and suicidal intent (AUC = 0·60). Cox-proportional hazards models suggested individuals utilizing tools with in-person care referral were significantly more likely to subsequently search for methods to actively end their life (HR = 1·727, p = 0·007). DiscussionOnline screens may influence help-seeking behavior, suicidal ideation, and suicidal intent. Websites with referrals to in-person treatments could put persons at greater risk of active suicidal intent. Further evaluation using large-scale randomized controlled trials is needed.

Using the internet search data to investigate symptom characteristics of COVID-19: A big data study.

ObjectiveAnalyzing the symptom characteristics of Coronavirus Disease 2019(COVID-19) to improve control and prevention.MethodsUsing the Baidu Index Platform (http://index.baidu.com) and the website of Chinese Center for Disease Control and Prevention as data resources to obtain the search volume (SV) of keywords for symptoms associated with COVID-19 from January 1 to February 20 in each year from 2017 to 2020 and the epidemic data in Hubei province and the other top 9 impacted provinces in China. Data of 2020 were compared with those of the previous three years. Data of Hubei province were compared with those of the other 9 provinces. The differences and characteristics of the SV of COVID-19-related symptoms, and the correlations between the SV of COVID-19 and the number of newly confirmed/suspected cases were analyzed. The lag effects were discussed.ResultsComparing the SV from January 1, 2020 to February 20, 2020 with those for the same period of the previous three years, Hubei's SV for cough, fever, diarrhea, chest tightness, dyspnea, and other symptoms were significantly increased. The total SV of lower respiratory symptoms was significantly higher than that of upper respiratory symptoms (P＜0.001). The SV of COVID-19 in Hubei province was significantly correlated with the number of newly confirmed/suspected cases (rconfirmed = 0.723, rsuspected = 0.863, both p < 0.001). The results of the distributed lag model suggested that the patients who searched relevant symptoms on the Internet may begin to see doctors in 2–3 days later and be confirmed in 3–4 days later.ConclusionThe total SV of lower respiratory symptoms was higher than that of upper respiratory symptoms, and the SV of diarrhea also increased significantly. It warned us to pay attention to not only the symptoms of the lower respiratory tract but also the gastrointestinal symptoms, especially diarrhea in patients with COVID-19. Internet search behavior had a positive correlation with the number of newly confirmed/suspected cases, suggesting that big data has an important role in the early warning of infectious diseases.

Chilling effects and the stock market response to the Snowden revelations

This study analyzes the changes in individuals’ Internet search behavior and economic variables following Edward Snowden’s revelation of National Security Agency (NSA) surveillance activities of global Internet communications. We refer to the abrupt changes in individual search behavior beginning in June 2013 following Snowden’s revelations as the “Snowden effect.” The “Snowden effect” is a form of chilling effect. Using data from a unique search engine, DuckDuckGo, we show that chilling effects indeed exist. However, further analysis including the stock prices of the largest cybersecurity companies demonstrate a muted market response to the revelations, while cybersecurity spending data suggest resulting economic harm in the form of increased opportunity costs of buyers of cybersecurity services.

Global Infodemiology of COVID-19: Analysis of Google Web Searches and Instagram Hashtags.

BackgroundAlthough “infodemiological” methods have been used in research on coronavirus disease (COVID-19), an examination of the extent of infodemic moniker (misinformation) use on the internet remains limited.ObjectiveThe aim of this paper is to investigate internet search behaviors related to COVID-19 and examine the circulation of infodemic monikers through two platforms—Google and Instagram—during the current global pandemic.MethodsWe have defined infodemic moniker as a term, query, hashtag, or phrase that generates or feeds fake news, misinterpretations, or discriminatory phenomena. Using Google Trends and Instagram hashtags, we explored internet search activities and behaviors related to the COVID-19 pandemic from February 20, 2020, to May 6, 2020. We investigated the names used to identify the virus, health and risk perception, life during the lockdown, and information related to the adoption of COVID-19 infodemic monikers. We computed the average peak volume with a 95% CI for the monikers.ResultsThe top six COVID-19–related terms searched in Google were “coronavirus,” “corona,” “COVID,” “virus,” “corona virus,” and “COVID-19.” Countries with a higher number of COVID-19 cases had a higher number of COVID-19 queries on Google. The monikers “coronavirus ozone,” “coronavirus laboratory,” “coronavirus 5G,” “coronavirus conspiracy,” and “coronavirus bill gates” were widely circulated on the internet. Searches on “tips and cures” for COVID-19 spiked in relation to the US president speculating about a “miracle cure” and suggesting an injection of disinfectant to treat the virus. Around two thirds (n=48,700,000, 66.1%) of Instagram users used the hashtags “COVID-19” and “coronavirus” to disperse virus-related information.ConclusionsGlobally, there is a growing interest in COVID-19, and numerous infodemic monikers continue to circulate on the internet. Based on our findings, we hope to encourage mass media regulators and health organizers to be vigilant and diminish the use and circulation of these infodemic monikers to decrease the spread of misinformation.

Understanding Facilitators and Barriers of Online Cancer Information Utilization among Cancer Survivors and their Families: Focus on the Theory of Planned Behavior.

Objective:Cancer survivors have various health care needs and are willing to be proactive with their health maintenance. Online information would be a useful resource to guide cancer survivors and their family members. Therefore, identifying the factors that influence Internet searching behaviors among cancer survivors and their family members is a first step toward providing better health care services for cancer care. Methods:We performed focus group interviews that were based on the Theory of Planned Behavior, with thirty-one participants to explore factors related to Internet search behaviors among cancer survivors and their family members. Results:Six themes were identified in the analysis of participant interviews. Attitudes toward searching for health information on the Internet included the themes “Fulfilling unmet needs” and “Confirmation through second opinion.” Themes related to social norms included “a required step for sure” and “helping each other.” In terms of perceived behavioral control, themes included “difficult to choose because of being ‘overwhelmed with information,’” and “complex searching milieu.” Conclusion:It was clear that cancer survivors and their family members had unmet needs for maintaining their health status. They wanted to be informed and actively involved in the decision-making process regarding health management. Consultation and education provided to patients by doctors should not only include information on diet and nutrition but also information on the resulting complications to satisfy their need for reliable health information.

Google Searches and Suicide Rates in Spain, 2004-2013: Correlation Study.

BackgroundDifferent studies have suggested that web search data are useful in forecasting several phenomena from the field of economics to epidemiology or health issues.ObjectiveThis study aimed to (1) evaluate the correlation between suicide rates released by the Spanish National Statistics Institute (INE) and internet search trends in Spain reported by Google Trends (GT) for 57 suicide-related terms representing major known risks of suicide and an analysis of these results using a linear regression model and (2) study the differential association between male and female suicide rates published by the INE and internet searches of these 57 terms.MethodsThe study period was from 2004 to 2013. In this study, suicide data were collected from (1) Spain’s INE and (2) local internet search data from GT, both from January 2004 to December 2013. We investigated and validated 57 suicide-related terms already tested in scientific studies before 2015 that would be the best predictors of new suicide cases. We then evaluated the nowcasting effects of a GT search through a cross-correlation analysis and by linear regression of the suicide incidence data with the GT data.ResultsSuicide rates in Spain in the study period were positively associated (r<-0.2) for the general population with the search volume for 7 terms and negatively for 1 from the 57 terms used in previous studies. Suicide rates for men were found to be significantly different than those of women. The search term, “allergy,” demonstrated a lead effect for new suicide cases (r=0.513; P=.001). The next significant correlating terms for those 57 studied were “antidepressant,” “alcohol abstinence,” “relationship breakup” (r=0.295, P=.001; r=0.295, P=.001; and r=0.268, P=.002, respectively). Significantly different results were obtained for men and women. Search terms that correlate with suicide rates of women are consistent with previous studies, showing that the incidence of depression is higher in women than in men, and showing different gender searching patterns.ConclusionsA better understanding of internet search behavior of both men and women in relation to suicide and related topics may help design effective suicide prevention programs based on information provided by search robots and other big data sources.

Detecting Lung Cancer Trends by Leveraging Real-World and Internet-Based Data: Infodemiology Study.

BackgroundInternet search data on health-related terms can reflect people’s concerns about their health status in near real time, and hence serve as a supplementary metric of disease characteristics. However, studies using internet search data to monitor and predict chronic diseases at a geographically finer state-level scale are sparse.ObjectiveThe aim of this study was to explore the associations of internet search volumes for lung cancer with published cancer incidence and mortality data in the United States.MethodsWe used Google relative search volumes, which represent the search frequency of specific search terms in Google. We performed cross-sectional analyses of the original and disease metrics at both national and state levels. A smoothed time series of relative search volumes was created to eliminate the effects of irregular changes on the search frequencies and obtain the long-term trends of search volumes for lung cancer at both the national and state levels. We also performed analyses of decomposed Google relative search volume data and disease metrics at the national and state levels.ResultsThe monthly trends of lung cancer-related internet hits were consistent with the trends of reported lung cancer rates at the national level. Ohio had the highest frequency for lung cancer-related search terms. At the state level, the relative search volume was significantly correlated with lung cancer incidence rates in 42 states, with correlation coefficients ranging from 0.58 in Virginia to 0.94 in Oregon. Relative search volume was also significantly correlated with mortality in 47 states, with correlation coefficients ranging from 0.58 in Oklahoma to 0.94 in North Carolina. Both the incidence and mortality rates of lung cancer were correlated with decomposed relative search volumes in all states excluding Vermont.ConclusionsInternet search behaviors could reflect public awareness of lung cancer. Research on internet search behaviors could be a novel and timely approach to monitor and estimate the prevalence, incidence, and mortality rates of a broader range of cancers and even more health issues.

Does Investor Attention to Accounting Information Matter to the Pricing of Stocks?

Based on Internet search behavior, I create a novel measure of individual investor attention to firm-specific financial information. To do this, I use search terms that combine a word that indicates the firm’s identity with a word that indicates the type of information the investor intends to pay attention to. When individual investor attention to financial information is high, I find that responses to earnings announcements are stronger, a finding that does not hold for search volume. Congruently, I find that the post-earnings announcement drift and the underreaction to earnings are weaker. However, I find a stronger overreaction to accruals, which is a less persistent component of earnings. The last result is unique to the active attention of individual investors and does not hold for sophisticated investors or for passive media-driven attention. These findings indicate that individual investors’ limitations in processing nuanced financial information are not fully offset by paying attention.

Infodemiology of Crohn's disease and Ulcerative colitis using GoogleTrends - an approach to investigate patient needs.

Inflammatory bowel diseases (IBD) highly affect quality of life. The course of disease and success of treatment are variable. These factors result in a high number of psychiatric comorbidities with patients requiring extensive medical consultation or additional psychotherapy. Unfortunately, time is often limited in daily clinical care, which leaves patients not feeling sufficiently informed about their disease. Patients often compensate by searching the internet, with possibly harmful information. We aim to identify information gaps to allow a more complete education of patients. We analyzed the internet search behavior using the key words [Morbus Crohn] (CD) and [Colitis ulcerosa] (UC) using Google Trends. In addition, we investigated which websites are the first hits on Google, as those are most likely visited by patients. Symptoms, nutrition and therapy are central topics for persons interested in IBD. The searches concerning symptoms or therapies do not match the actual incidence or prevalence of comorbidities as well as the more commonly used therapies or established nutritional recommendations. We found a distinct impact of well-known personalities on disease related searches. The first suggested websites on google showed great heterogeneity of responsible publishers, often with possible conflict of interests. In line with those observations, quality of website content is highly heterogeneous. This study showed a need of information concerning symptoms, nutrition and therapy that should be considered during patient education. Since time in physician patient dialogue is short it may be helpful to further evaluate websites independently in order to give recommendations of websites offering reliable information.

DISCOVER POINTS OF INTEREST BASED ON USERS’ INTERNET SEARCHES THROUGH AN ONLINE SHOPPING WEBSITE

Abstract. In recent years, the development of the Internet plays a significant role in human's daily activities. One of the most important effects of the Internet is the change in the process of shopping. The advent of online shopping leads to establish a new channel for customers to obtain information about their desired goods and demands. Although many customers collect information from online channel, they also wish to try and search for their required goods at the stores. Besides, discovering this data leads to a new source for spatial analysis to find the users’ interests. Therefore, we can consider this data as a contextual information source for spatial analysis or primary source for recommending points of interest (POIs). In this research, our aim is to discover a relation among the users' internet searches and the goods at the stores to recommend the best store to the users. Euclidean distance is used to calculate the similarity between users' searches and the available goods at the stores. The proposed method has been implemented in the city of Tehran, capital of Iran. The results show that the users’ internet search behavior plays an essential role in the recommendation system which provides stores to the users based on the similarity among the users’ internet searches and the available goods at the stores.

Online Search Behavior for Cancer Immunotherapy Resources and Readability Analysis: An Opportunity to Aid in Medical Decision-making.

Cancer patients are faced with increasing options for cancer care, especially with the introduction of cancer immunotherapy with immune checkpoint inhibitors (ICIs). Though many patients turn to online resources to supplement their decision-making, it is unknown whether online resources in cancer immunotherapy with ICIs are written at an appropriate level of readability according to national medical organizations. We performed a cross-sectional analysis of internet search behavior for cancer immunotherapy by ICIs and clinical trial availability per ClinicalTrials.gov in the United States (US) from 2004 - 2018 with subsequent quantitation of readability by four readability formulas of top 50 online resources. Internet search behavior for “cancer immunotherapy” has steadily increased since 2013 and coincides with the year of the US Food and Drug Administration (FDA) approval for individual ICIs. Furthermore, internet search behavior was significantly correlated with clinical trial availability in the US (R = 0.97, p < 0.0001). None of the top 50 resources available to patients were found to be within the recommended level of sixth-grade readability or less with only one (2%) written at the middle school level, 21 (42%) at the high school level, 23 (46%) at the university level, and five (8%) at a graduate level. Population-level internet search patterns may reflect patient behavior in seeking relevant online health information and may be influenced by new options for cancer therapy, including via clinical trials. However, low readability of available online resources may impede patient comprehension and negatively affect medical decision-making.

Bargaining in the Shadow of the Trial? Deaths of Law Enforcement Officials and the Plea Bargaining Process

Plea is the cornerstone of the U.S. criminal justice system and the in the shadow of the framework, where the plea reached is driven primarily by the expected sentence arising from a trial, is the convention for applied economists. Criminologists and legal scholars challenge the bargaining in the shadow of the trial model's accuracy. There has not been a test of the validity of the conceptual framework. We provide one. We use a large data set of felony cases in Florida to estimate the plea discount received. Our novel identification strategy is to consider deaths of law enforcement officials, which we argue is a newsworthy, tragic event affecting a local community and making violent crime salient to the citizens who make up the potential jury pool. Those cases, unrelated to the death, but already in process at the time and in the same location as the death, act as our treated observations experiencing an exogenous shock to their probability of conviction. We show that these individuals plea guilty to substantially longer sentences but, importantly, that the difference in the sanctions between guilty pleas and convictions reduces. This effect is especially strong for deaths of law enforcement officials who die via gunfire, when there is more internet search behavior by the local population, when the death occurs closer to the date, and when road dedications were made in the fallen officers names. The reduction in the plea discount is greater for serious crimes, but also exists for less-serious crimes. Thus, plea becomes relatively less generous. Theory does not predict, though, what will happen to the rate since tougher offers from the prosecutor should lead to more trials, but the heightened conviction probability should encourage negotiation. We find that the likelihood of a increases, consistent with the hypothesis that prosecutors are making less-generous plea offers. Thus, we provide strong evidence that plea occurs in the shadow of the trial.

Can search engine data improve accuracy of demand forecasting for new products? Evidence from automotive market

Purpose The purpose of this paper is to analyze the relationship between new product diffusion and consumer internet search patterns using big data and to investigate whether such data can be used in forecasting new product diffusion. Design/methodology/approach This research proposes a new product diffusion model based on the Bass diffusion model by incorporating consumer internet search behavior. Actual data from search engine queries and new vehicle sales for each vehicle class and region are used to estimate the proposed model. Statistical analyses are used to interpret the estimated results, and the prediction performance of the proposed method is compared with other methods to validate the usefulness of data for internet search engine queries in forecasting new product diffusion. Findings The estimated coefficients of the proposed model provide a clear interpretation of the relationship between new product diffusion and internet search volume. In 83.62 percent of 218 cases, analyzing the internet search pattern data are significant to explain new product diffusion and that internet search volume helps to predict new product diffusion. Therefore, marketing that seeks to increase internet search volume could positively affect vehicle sales. In addition, the demand forecasting performance of the proposed diffusion model is superior to those of other models for both long-term and short-term predictions. Research limitations/implications As search queries have only been available since 2004, comparisons with data from earlier years are not possible. The proposed model can be extended using other big data from additional sources. Originality/value This research directly demonstrates the relationship between new product diffusion and consumer internet search pattern and investigates whether internet search queries can be used to forecast new product diffusion by product type and region. Based on the estimated results, increasing internet search volume could positively affect vehicle sales across product types and regions. Because the proposed model had the best prediction power compared with the other considered models for all cases with large margins, it can be successfully utilized in forecasting demand for new products.

Correlation of Lung Cancer Incidence and Mortality with Internet Search Data in US States: 2004 to 2018

Background: Internet search data reflect population's concern on health topics in near-real time and can serve as an supplementary metric of disease characteristics. We present a detailed analysis of the association among Google relative search volumes (RSV), lung cancer incidence and mortality rates in the United State. We forecast the trends of lung cancer on this basis. Methods: We obtained RSV of lung cancer from Google Trends. We collected incidence and mortality rates from the US Centers for Disease Control and Prevention. We further created a smoothed time series of RSV so as to eliminate the effects of irregular changes and obtain the long-term trend. We performed a cross-sectional analysis of original Google RSV data and disease metrics at the national and state level. We also performed a analysis of the decomposed RSV and disease metrics. Findings: Trend of lung cancer related internet searches is consistent with the trend of lung cancer incidence nationally. Ohio had the highest RSV. At the state level, the RSV is statistically significantly correlated with lung cancer incidence in 42 regions, with correlation ranging from 0.578 to 0.943. RSV is also statistically significantly correlated with mortality in 47 regions. Both the lung cancer incidence and mortality were correlated with decomposed RSV in 50 regions, except Vermont. Interpretation: Internet searches indeed reflect public awareness of cancer. Research on internet search behaviors could offer an innovative and timely approach to monitoring and estimating lung cancer incidence and mortality. Funding Statement: This research was funded by National Natural Science Foundation of China (#91746205). Declaration of Interests: The authors state: None declared. Ethics Approval Statement: Not required.

Association Between Cancer Incidence and Mortality in Web-Based Data in China: Infodemiology Study.

BackgroundCancer poses a serious threat to the health of Chinese people, resulting in a major challenge for public health work. Today, people can obtain relevant information from not only medical workers in hospitals, but also the internet in any place in real-time. Search behaviors can reflect a population’s awareness of cancer from a completely new perspective, which could be driven by the underlying cancer epidemiology. However, such Web-retrieved data are not yet well validated or understood.ObjectiveThis study aimed to explore whether a correlation exists between the incidence and mortality of cancers and normalized internet search volumes on the big data platform, Baidu. We also assessed whether the distribution of people who searched for specific types of cancer differed by gender. Finally, we determined whether there were regional disparities among people who searched the Web for cancer-related information.MethodsStandard Boolean operators were used to choose search terms for each type of cancer. Spearman’s correlation analysis was used to explore correlations among monthly search index values for each cancer type and their monthly incidence and mortality rates. We conducted cointegration analysis between search index data and incidence rates to examine whether a stable equilibrium existed between them. We also conducted cointegration analysis between search index data and mortality data.ResultsThe monthly Baidu index was significantly correlated with cancer incidence rates for 26 of 28 cancers in China (lung cancer: r=.80, P<.001; liver cancer: r=.28, P=.016; stomach cancer: r=.50, P<.001; esophageal cancer: r=.50, P<.001; colorectal cancer: r=.81, P<.001; pancreatic cancer: r=.86, P<.001; breast cancer: r=.56, P<.001; brain and nervous system cancer: r=.63, P<.001; leukemia: r=.75, P<.001; Non-Hodgkin lymphoma: r=.88, P<.001; Hodgkin lymphoma: r=.91, P<.001; cervical cancer: r=.64, P<.001; prostate cancer: r=.67, P<.001; bladder cancer: r=.62, P<.001; gallbladder and biliary tract cancer: r=.88, P<.001; lip and oral cavity cancer: r=.88, P<.001; ovarian cancer: r=.58, P<.001; larynx cancer: r=.82, P<.001; kidney cancer: r=.73, P<.001; squamous cell carcinoma: r=.94, P<.001; multiple myeloma: r=.84, P<.001; thyroid cancer: r=.77, P<.001; malignant skin melanoma: r=.55, P<.001; mesothelioma: r=.79, P<.001; testicular cancer: r=.57, P<.001; basal cell carcinoma: r=.83, P<.001). The monthly Baidu index was significantly correlated with cancer mortality rates for 24 of 27 cancers. In terms of the whole population, the number of women who searched for cancer-related information has slowly risen over time. People aged 30-39 years were most likely to use search engines to retrieve cancer-related knowledge. East China had the highest Web search volumes for cancer.ConclusionsSearch behaviors indeed reflect public awareness of cancer from a different angle. Research on internet search behaviors could present an innovative and timely way to monitor and estimate cancer incidence and mortality rates, especially for cancers not included in national registries.