A report published in the American Journal of Toxicology details how medical researchers are searching for users on Twitter who are engaging in prescription abuse.
According to the report, which was published online in August, the purpose of the study was to demonstrate that there was a strong correlation between the geographic location that a post was made and estimates that the government has made regarding misuse of prescription opioids.
The report states that the goal was to ascertain whether or not Twitter could be used as a longitudinal source of data on the opioid addiction epidemic in the U.S.
In order to gather the data, researchers used Twitter’s application programming interface (API). The API allows for access to all posts on Twitter that are publically available and delineated by any filter desired.
Researchers wrote a program that identified tweets that had certain keywords within them. These keywords included slang related to opioids and prescription abuse. The tweets were then rated as “not related” or “related” to the misuse of prescription opioids based on their context, thus eliminating tweets which might seem to contain related keywords, but which were actually about other topics or legitimate opioid use.
Roughly 2% of tweets contain a precise geographic location. Using an algorithm, researchers were able to use metadata to estimate the geographic origin of the post about opioid prescription abuse with greater than 80% probability.
Ultimately, the researchers found that their estimates of rates of prescription abuse in particular geographic locations agreed with government-gathered national survey data.
Researchers believe that data gathered from publicly-available twitter posts can prove useful in gathering epidemiological data about the opioid prescription abuse epidemic, as well as providing an expedient and swift method of gathering information regarding the geographic spread of substance abuse.