Artificial intelligence and venereal diseases

The Internet gives humanity unprecedented opportunities to collect a large amount of statistical information about people. Search engines and social networks can calculate the number of requests and mentions of different things. This has long been understood in the advertising industry and with might and main use personal preferences and search history. Develop this niche and scientists, studies conducted at the University of California sought a link between the various search queries related and risky secular behavior and the number of sexually transmitted diseases. The connection was found, it turned out that the number of patients with syphilis can be predicted in advance.

Two studies were conducted in collaboration with the Centers for Disease Control and Prevention (CDC), analyzing the behavior of users in the Google search engine and in a social network with short Twitter messages. After that, medical data were taken on syphilis diseases by district and state, and all these data sets were compared to identify trends.

Sean Young, the founder and director of the Center for Digital Behavior, said that many of today's health problems can be tackled by seeing trends and taking action ahead of time. Venereal diseases, HIV, syphilis and other sexually transmitted diseases, drug addiction, and even cancer - to choose the best strategies to fight them you need to know not only the current state, but also to predict the situation in the future, to see trends.

During the work with syphilis, scientists developed a methodology for using search data for similar studies on medical topics in the future.

In the first part, scientists collected search queries in Google for each of the US states and compared them to registered cases of syphilis - primary and secondary - these are the earliest and most contagious stages of a sexually transmitted infection. Such information was provided by state bodies - centers for the control and prevention of diseases.In total, 25 key words and phrases (such as "find sex" were chosen, although the Russian translation would probably be worse, "sexually transmitted diseases," etc.), entered by users in the Google search system from January 1 2012 to December 31, 2014. After that, they took the CDC statistics on the weekly levels of treatment of patients with syphilis in the districts of all fifty US states.

Further, these data were compared using a static computer model, with machine learning, which scans large amounts of data and finds dependencies. This algorithm, based on artificial intelligence, was used first for training, and then for predicting the number of diseases by search queries with keywords.

The researchers found that the model predicts the level of disease for each state with 90 percent accuracy, which was an excellent result.

In the second part of the work done the same thing with user messages on Twitter. The information was collected for the period from May 26 to December 9, 2012 - there were 8,538 messages with keywords that contain geographic tags, allowing to identify the location (staff) of the person who wrote the tweet.

As in the first case, the algorithm tried to find dependencies with the number of cases of patients with syphilis. And also the connection of risky sexual behavior on the Internet and the increase in the number of diseases was found. In addition, the amount of another stage of the disease - early latent syphilis, that is detected only a year after infection, was investigated. And the data for 2013 also repeated search requests of a year ago.

States and counties, where Twitter users behaved frivolously, showed an increase in primary-stage diseases by 2.7 percent, secondary-by 3.6 percent.As a result, scientists have shown by their example that having inexpensive and large-scale statistical data, one can predict incidence trends in different states and districts, organize measures to prevent the development of epidemics, etc. Such algorithms can be used by other researchers, in other fields of science, in other countries.