Predicting Smoking Prevalence in Japan Using Search Volumes in an Internet Search Engine: Infodemiology Study.

BACKGROUND: Tobacco smoking is an important public health issue and a core indicator of public health policy worldwide. However, global pandemics and natural disasters have prevented surveys from being conducted. OBJECTIVE: The purpose of this study was to predict smoking prevalence by prefecture and sex in Japan using Internet search trends. METHODS: This study used the infodemiology approach. The outcome variable was smoking prevalence by prefecture, obtained from national surveys. The predictor variables were the search volumes on Yahoo! Japan Search. We collected the search volumes for queries related to terms from the thesaurus of the Japanese medical article database Ichu-shi. Predictor variables were converted to per capita values and standardized as z scores. For smoking prevalence, the values for 2016 and 2019 were used, and for search volume, the values for the April 1 to March 31 fiscal year (FY) 1 year prior to the survey (ie, FY 2015 and FY 2018) were used. Partial correlation coefficients, adjusted for data year, were calculated between smoking prevalence and search volume, and a regression analysis using a generalized linear mixed model with random effects was conducted for each prefecture. Several models were tested, including a model that included all search queries, a variable reduction method, and one that excluded cigarette product names. The best model was selected with the Akaike information criterion corrected (AICC) for small sample size and the Bayesian information criterion (BIC). We compared the predicted and actual smoking prevalence in 2016 and 2019 based on the best model and predicted the smoking prevalence in 2022. RESULTS: The partial correlation coefficients for men showed that 9 search queries had significant correlations with smoking prevalence, including cigarette (r=-0.417, P<.001), cigar in kanji (r=-0.412, P<.001), and cigar in katakana (r=-0.399, P<.001). For women, five search queries had significant correlations, including vape (r=0.335, P=.001), quitting smoking (r=0.288, P=.005), and cigar (r=0.286, P=.006). The models with all search queries were the best models for both AICC and BIC scores. Scatter plots of actual and estimated smoking prevalence in 2016 and 2019 confirmed a relatively high degree of agreement. The average estimated smoking prevalence in 2022 in the 47 prefectures for the total sample was 23.492% (95% CI 21.617%-25.367%), showing an increasing trend, with an average of 29.024% (95% CI 27.218%-30.830%) for men and 8.793% (95% CI 7.531%-10.054%) for women. CONCLUSIONS: This study suggests that the search volume of tobacco-related queries in internet search engines can predict smoking prevalence by prefecture and sex in Japan. These findings will enable the development of low-cost, timely, and crisis-resistant health indicators that will enable the evaluation of health measures and contribute to improved public health.

Measuring concerns about the COVID-19 vaccine among Japanese internet users through search queries.

With the increasing availability of the COVID-19 vaccines, vaccination has been rapidly promoted globally as a countermeasure against the spread of COVID-19. In Japan, vaccination was first introduced in February 2021. However, the amount of concern towards vaccination differs between individuals, and topics of concern include adverse reactions and side effects. This study investigated attitudes toward vaccines or vaccination during the COVID-19 pandemic across different Japanese prefectures, using Yahoo! JAPAN search queries. We first defined a vaccine concern index (VCI) by aggregating the search counts of vaccine-related queries from Yahoo! JAPAN users before examining VCI across all Japanese prefectures, accounting for gender and age. Our results demonstrated that VCI tended to be lower in more populated areas, and VCI was higher in their 20s to 40s than older people, especially in female users. Furthermore, there was a significant positive correlation (Spearman's Rank correlation coefficient [Formula: see text] = 0.60, [Formula: see text]) between VCI and prefectural vaccination rate, suggesting that web searching of adverse vaccine reactions may precede actual vaccination. This could reflect the information-seeking behavior of individuals who are accepting of vaccinations.

Prediction of Age-Adjusted Mortality From Stroke in Japanese Prefectures: Ecological Study Using Search Engine Queries.

BACKGROUND: Stroke is a major cause of death and the need for nursing care in Japan, with large regional disparities. OBJECTIVE: The purpose of this study was to clarify the association between stroke-related information retrieval behavior and age-adjusted mortality in each prefecture in Japan. METHODS: Age-adjusted mortality from stroke and aging rates were obtained from publicly available Japanese government statistics. A total of 9476 abstracts of Japanese articles related to symptoms and signs of stroke were identified in Ichushi-Web, a Japanese web-based database of biomedical articles, and 100 highly frequent words (hereafter referred to as the Stroke 100) were extracted. Using data from 2014 to 2019, a random forest analysis was carried out using the age-adjusted mortality from stroke in 47 prefectures as the outcome variable and the standardized retrieval numbers of the Stroke 100 words in the log data of Yahoo! JAPAN Search as predictive variables. Regression analysis was performed using a generalized linear mixed model (GLMM) with the number of standardized searches for Stroke 100 words with high importance scores in the random forest model as the predictive variable. In the GLMM, the aging rate and data year were used as control variables, and the random slope of data year and random intercept were calculated by prefecture. RESULTS: The mean age-adjusted mortality from stroke was 28.07 (SD 4.55) deaths per 100,000 for all prefectures in all data years. The accuracy score of the random forest analysis was 89.94%, the average error was 2.79 degrees, and the mean squared error was 13.57 degrees. The following 9 variables with high importance scores in the random forest analysis were selected as predictive variables for the regression analysis: male, age, hospitalization, enforcement, progress, stroke, abnormal, use, and change. As a result of the regression analysis with GLMM, the standardized partial regression coefficients (ß) and 95% confidence intervals showed that the following internet search terms were significantly associated with age-adjusted mortality from stroke: male (ß=-5.83, 95% CI -8.67 to -3.29), age (ß=-5.83, 95% CI -8.67 to -3.29), hospitalization (ß=-5.83, 95% CI -8.67 to -3.29), and abnormal (ß=3.83, 95% CI 1.14 to 6.56). CONCLUSIONS: Stroke-related search behavior was associated with age-adjusted mortality from stroke in each prefecture in Japan. Query terms that were strongly associated with age-adjusted mortality rates of stroke suggested the possibility that individual characteristics, such as sex and age, have an impact on stroke-associated mortality and that it is important to receive medical care early after stroke onset. Further studies on the criteria and timing of alerting are needed by monitoring information-seeking behavior to identify queries that are strongly associated with stroke mortality.

Predicting the Number of Suicides in Japan Using Internet Search Queries: Vector Autoregression Time Series Model.

BACKGROUND: The number of suicides in Japan increased during the COVID-19 pandemic. Predicting the number of suicides is important to take timely preventive measures. OBJECTIVE: This study aims to clarify whether the number of suicides can be predicted by suicide-related search queries used before searching for the keyword "suicide." METHODS: This study uses the infoveillance approach for suicide in Japan by search trends in search engines. The monthly number of suicides by gender, collected and published by the National Police Agency, was used as an outcome variable. The number of searches by gender with queries associated with "suicide" on "Yahoo! JAPAN Search" from January 2016 to December 2020 was used as a predictive variable. The following five phrases highly relevant to suicide were used as search terms before searching for the keyword "suicide" and extracted and used for analyses: "abuse"; "work, don't want to go"; "company, want to quit"; "divorce"; and "no money." The augmented Dickey-Fuller and Johansen tests were performed for the original series and to verify the existence of unit roots and cointegration for each variable, respectively. The vector autoregression model was applied to predict the number of suicides. The Breusch-Godfrey Lagrangian multiplier (BG-LM) test, autoregressive conditional heteroskedasticity Lagrangian multiplier (ARCH-LM) test, and Jarque-Bera (JB) test were used to confirm model convergence. In addition, a Granger causality test was performed for each predictive variable. RESULTS: In the original series, unit roots were found in the trend model, whereas in the first-order difference series, both men (minimum tau 3: -9.24; max tau 3: -5.38) and women (minimum tau 3: -9.24; max tau 3: -5.38) had no unit roots for all variables. In the Johansen test, a cointegration relationship was observed among several variables. The queries used in the converged models were "divorce" for men (BG-LM test: P=.55; ARCH-LM test: P=.63; JB test: P=.66) and "no money" for women (BG-LM test: P=.17; ARCH-LM test: P=.15; JB test: P=.10). In the Granger causality test for each variable, "divorce" was significant for both men (F104=3.29; P=.04) and women (F104=3.23; P=.04). CONCLUSIONS: The number of suicides can be predicted by search queries related to the keyword "suicide." Previous studies have reported that financial poverty and divorce are associated with suicide. The results of this study, in which search queries on "no money" and "divorce" predicted suicide, support the findings of previous studies. Further research on the economic poverty of women and those with complex problems is necessary.

Measuring Public Concern About COVID-19 in Japanese Internet Users Through Search Queries: Infodemiological Study.

BACKGROUND: COVID-19 has disrupted lives and livelihoods and caused widespread panic worldwide. Emerging reports suggest that people living in rural areas in some countries are more susceptible to COVID-19. However, there is a lack of quantitative evidence that can shed light on whether residents of rural areas are more concerned about COVID-19 than residents of urban areas. OBJECTIVE: This infodemiology study investigated attitudes toward COVID-19 in different Japanese prefectures by aggregating and analyzing Yahoo! JAPAN search queries. METHODS: We measured COVID-19 concerns in each Japanese prefecture by aggregating search counts of COVID-19-related queries of Yahoo! JAPAN users and data related to COVID-19 cases. We then defined two indices-the localized concern index (LCI) and localized concern index by patient percentage (LCIPP)-to quantitatively represent the degree of concern. To investigate the impact of emergency declarations on people's concerns, we divided our study period into three phases according to the timing of the state of emergency in Japan: before, during, and after. In addition, we evaluated the relationship between the LCI and LCIPP in different prefectures by correlating them with prefecture-level indicators of urbanization. RESULTS: Our results demonstrated that the concerns about COVID-19 in the prefectures changed in accordance with the declaration of the state of emergency. The correlation analyses also indicated that the differentiated types of public concern measured by the LCI and LCIPP reflect the prefectures' level of urbanization to a certain extent (ie, the LCI appears to be more suitable for quantifying COVID-19 concern in urban areas, while the LCIPP seems to be more appropriate for rural areas). CONCLUSIONS: We quantitatively defined Japanese Yahoo users' concerns about COVID-19 by using the search counts of COVID-19-related search queries. Our results also showed that the LCI and LCIPP have external validity.

Predicting regional influenza epidemics with uncertainty estimation using commuting data in Japan.

Obtaining an accurate prediction of the number of influenza patients in specific areas is a crucial task undertaken by medical institutions. Infections (such as influenza) spread from person to person, and people are rarely confined to a single area. Therefore, creating a regional influenza prediction model should consider the flow of people between different areas. Although various regional flu prediction models have previously been proposed, they do not consider the flow of people among areas. In this study, we propose a method that can predict the geographical distribution of influenza patients using commuting data to represent the flow of people. To elucidate the complex spatial dependence relations, our model uses an extension of the graph convolutional network (GCN). Additionally, a prediction interval for medical institutions is proposed, which is suitable for cyclic time series. Subsequently, we used the weekly data of flu patients from health authorities as the ground-truth to evaluate the prediction interval and performance of influenza patient prediction in each prefecture in Japan. The results indicate that our GCN-based model, which used commuting data, considerably improved the predictive accuracy over baseline values both temporally and spatially to provide an appropriate prediction interval. The proposed model is vital in practical settings, such as in the decision making of public health authorities and addressing growth in vaccine demand and workload. This paper primarily presents a GCN as a useful means for predicting the spread of an epidemic.

Surveillance of early stage COVID-19 clusters using search query logs and mobile device-based location information.

Two clusters of the coronavirus disease 2019 (COVID-19) were confirmed in Hokkaido, Japan, in February 2020. To identify these clusters, this study employed web search query logs of multiple devices and user location information from location-aware mobile devices. We anonymously identified users who used a web search engine (i.e., Yahoo! JAPAN) to search for COVID-19 or its symptoms. We regarded them as web searchers who were suspicious of their own COVID-19 infection (WSSCI). We extracted the location of WSSCI via a mobile operating system application and compared the spatio-temporal distribution of WSSCI with the actual location of the two known clusters. In the early stage of cluster development, we confirmed several WSSCI. Our approach was accurate in this stage and became biased after a public announcement of the cluster development. When other cluster-related resources, such as detailed population statistics, are not available, the proposed metric can capture hints of emerging clusters.

Robust two-stage influenza prediction model considering regular and irregular trends.

Influenza causes numerous deaths worldwide every year. Predicting the number of influenza patients is an important task for medical institutions. Two types of data regarding influenza-like illnesses (ILIs) are often used for flu prediction: (1) historical data and (2) user generated content (UGC) data on the web such as search queries and tweets. Historical data have an advantage against the normal state but show disadvantages against irregular phenomena. In contrast, UGC data are advantageous for irregular phenomena. So far, no effective model providing the benefits of both types of data has been devised. This study proposes a novel model, designated the two-stage model, which combines both historical and UGC data. The basic idea is, first, basic regular trends are estimated using the historical data-based model, and then, irregular trends are predicted by the UGC data-based model. Our approach is practically useful because we can train models separately. Thus, if a UGC provider changes the service, our model could produce better performance because the first part of the model is still stable. Experiments on the US and Japan datasets demonstrated the basic feasibility of the proposed approach. In the dropout (pseudo-noise) test that assumes a UGC service would change, the proposed method also showed robustness against outliers. The proposed model is suitable for prediction of seasonal flu.

Comparing Medical Term Usage Patterns of Professionals and Search Engine and Community Question Answering Service Users in Japan: Log Analysis.

BACKGROUND: Despite increasing opportunities for acquiring health information online, discussion of the specific words used in searches has been limited. OBJECTIVE: The aim of this study was to clarify the medical information gap between medical professionals and the general public in Japan through health information-seeking activities on the internet. METHODS: Search and posting data were analyzed from one of the most popular domestic search engines in Japan (Yahoo! JAPAN Search) and the most popular Japanese community question answering service (Yahoo! Chiebukuro). We compared the frequency of 100 clinical words appearing in the clinical case reports of medical professionals (clinical frequency) with their frequency in Yahoo! JAPAN Search (search frequency) logs and questions posted to Yahoo! Chiebukuro (question frequency). The Spearman correlation coefficient was used to quantify association patterns among the three information sources. Additionally, user information (gender and age) in the search frequency associated with each registered user was extracted. RESULTS: Significant correlations were observed between clinical and search frequencies (r=0.29, P=.003), clinical and question frequencies (r=0.34, P=.001), and search and question frequencies (r=0.57, P<.001). Low-frequency words in clinical frequency (eg, "hypothyroidism," "ulcerative colitis") highly ranked in search frequency. Similarly, "pain," "slight fever," and "numbness" were highly ranked only in question frequency. The weighted average of ages was 34.5 (SD 2.7) years, and the weighted average of gender (man -1, woman +1) was 0.1 (SD 0.1) in search frequency. Some words were specifically extracted from the search frequency of certain age groups, including "abdominal pain" (10-20 years), "plasma cells" and "inflammatory findings" (20-30 years), "DM" (diabetes mellitus; 30-40 years), "abnormal shadow" and "inflammatory findings" (40-50 years), "hypertension" and "abnormal shadow" (50-60 years), and "lung cancer" and "gastric cancer" (60-70 years). CONCLUSIONS: Search and question frequencies showed similar tendencies, whereas search and clinical frequencies showed discrepancy. Low-clinical frequency words related to diseases such as "hypothyroidism" and "ulcerative colitis" had high search frequencies, whereas those related to symptoms such as "pain," "slight fever," and "numbness" had high question frequencies. Moreover, high search frequency words included designated intractable diseases such as "ulcerative colitis," which has an incidence of less than 0.1% in the Japanese population. Therefore, it is generally worthwhile to pay attention not only to major diseases but also to minor diseases that users frequently seek information on, and more words will need to be analyzed in the future. Some characteristic words for certain age groups were observed (eg, 20-40 years: "cancer"; 40-60 years: diagnoses and diseases identified in health examinations; 60-70 years: diseases with late adulthood onset and "death"). Overall, this analysis demonstrates that medical professionals as information providers should be aware of clinical frequency, and medical information gaps between professionals and the general public should be bridged.