Online searches for information about risk factors for and symptoms of syphilis can alert public health authorities to emerging cases, according to findings published in Epidemiology.
Google search data have been collected to monitor and predict infectious disease patterns for a few years. Google Flu Trends and Google Dengue Trends offer “nowcasting” data on these infections to researchers for the world and by country.
Sean D. Young, PhD, from the University of California, Los Angeles, Institute for Prediction Technology and colleagues applied the strategy to track the number of people searching Google for information on syphilis compared with state surveillance of cases published in the Morbidity and Mortality Weekly Report.
Reported rates of primary and secondary syphilis have been rising in the United States since 2001, including a 19.4% increase in 2015 from the previous year. These statistics likely underestimate incidence because some infected individuals do not present to a healthcare provider. Mining Google and social media such as Twitter may reveal cases that might not be included in public health surveillance.
The Centers for Disease Control and Prevention has issued guidelines on best practices for using online search data to track the spread of sexually transmitted infections.
Google Trends provides search query data by state and for the entire United States. The new study investigated whether Google Trends data could predict the number of weekly syphilis case reports from individual states and whether search data from the first 10 weeks of a year could predict new cases of primary and secondary syphilis for the rest of the year.
The researchers used linear regression analysis and machine learning techniques to connect Google Trends and Centers for Disease Control and Prevention data. Specifically, they considered Google Trends data using 25 risk-related keywords for recent primary and secondary syphilis searches from 2012 to 2014 compared with cases reported weekly to the Centers for Disease Control and Prevention for the 50 states from that period.
The team connected 155 weeks of Google Trends data to syphilis data reported a week later, multiplied by 50 states, to yield 7750 data points. They also trained the algorithm for the first 10 weeks of each year. Then they validated the models for 2012 and 2013 for the following 52 weeks, and the 2014 model for the following 42 weeks.
The models accurately predicted 144 weeks of primary and secondary syphilis counts for each state “with high accuracy,” the researchers write, with an overall average R 2 of 0.9 and overall average root mean squared error of 4.9. Weekly primary and secondary syphilis counts by state ranged from 0 to 151.
The predictions were not as accurate for two of the New Year’s weeks, which did not happen for other holidays and may reflect less Internet time because of the socializing that comes during the last week of the year. The Google data can be used to predict cases for up to a year, the researchers found.
The investigators suggest that Google Trends data supplement existing case reporting, with the ability to predict changes in the epidemic and to guide where syphilis awareness programs might best be deployed. Tapping into social media and Internet searches to track syphilis may be especially helpful in guiding prevention and treatment programs in nations that do not have robust public health case reporting.
Limitations of the study include the basis of the training models on reported cases of primary and secondary syphilis that probably underestimated incidence. The strategy would also miss people who contract syphilis who do not use the Internet or Google. The approach also did not have the granularity to focus data on regions within states that might have the greatest need for prevention, monitoring, and treatment programs.
“We found an association between freely available internet search query data and weekly reported cases of primary and secondary syphilis. Results suggest the need for further exploration on whether and how internet search data can be integrated into public health monitoring systems for sexually transmitted infections,” the researchers conclude.
The researchers have disclosed no relevant financial relationships.
Epidemiology. Published online April 7, 2018.
For more news, join us on Facebook and Twitter
Tidak ada komentar:
Posting Komentar