Google Trends correlation and sensitivity for outbreaks of dengue and yellow fever in the state of São Paulo

Abstract

Objective To assess Google Trends accuracy for epidemiological surveillance of dengue and yellow fever, and to compare the incidence of these diseases with the popularity of its terms in the state of São Paulo. Methods Retrospective cohort. Google Trends survey results were compared to the actual incidence of diseases, obtained from Centro de Vigilância Epidemiológica “Prof. Alexandre Vranjac” , in São Paulo, Brazil, in periods between 2017 and 2019. The correlation was calculated by Pearson’s coefficient and cross-correlation function. The accuracy was analyzed by sensitivity and specificity values. Results There was a statistically significant correlation between the variables studied for both diseases, Pearson coefficient of 0.91 for dengue and 0.86 for yellow fever. Correlation with up to 4 weeks of anticipation for time series was identified. Sensitivity was 87% and 90%, and specificity 69% and 78% for dengue and yellow fever, respectively. Conclusion The incidence of dengue and yellow fever in the State of São Paulo showed a strong correlation with the popularity of its terms measured by Google Trends in weekly periods. Google Trends tool provided early warning, with high sensitivity, for the detection of outbreaks of these diseases.

Trending ticks: using Google Trends data to understand tickborne disease prevention

“…Ticks and the pathogens they carry can cause diseases that are a tangible part of people’s everyday lives, prompting them to seek information about associated risks actively. Individuals situated in high-risk environments or periods of vector-borne diseases are more inclined to conduct online searches to arm themselves with the necessary knowledge to mitigate these risks, resulting in an overall increase in search activity ( 31 , 32 ). It reaffirmed that individuals face an increased risk in the northeastern regions (or eastern areas for lone star ticks), particularly during the summer months ( 3 , 6 , 7 , 11 , 40 , 44 ).…”

“…In addition, online search behaviors related to vector-borne diseases occurring during specific periods can serve as an effective tool for predicting, controlling, preventing, and reporting epidemic cases. For example, the incidence of mosquito-borne diseases such as dengue and yellow fever in tropical regions is closely correlated with the popularity of related search terms on online search engines ( 31 , 32 ). Such as (1) symptoms associated with mosquito-borne diseases, (2) preventive measures, such as “mosquito repellent,” “mosquito net,” “vaccination,” (3) geographic locations or regions affected by mosquito-borne diseases, such as “tropical regions,” “subtropical climates,” (4) specific countries or cities where outbreaks occur, and (5) public health campaigns or initiatives related to mosquito-borne diseases, such as “mosquito control,” “vector control,” “public health interventions,” among others.…”

O que o Google Trends tem a dizer sobre a proporção de usuários de narguilé na população brasileira?

“…Google Trends e proporção de usuários de narguilé na população brasileira desencadear mudanças de comportamento dos fumantes. 19 No Brasil, foi utilizado para prever a incidência e surtos de doenças infecciosas [19][20][21] e analisar o impacto de campanhas para prevenção e controle do câncer.…”

What does Google Trends reveal about the proportion of waterpipe users in the Brazilian population?

“…18 Furthermore, Google Trends has been used to understand how unforeseen acute events, such as the COVID-19 epidemic, can trigger changes in smokers' behavior. 19 In Brazil, it has been used to predict the incidence and outbreaks of infectious diseases [19][20][21] and to analyze the impact…”