Posts on publicly available social media platforms such as Twitter contain a huge volume of information about the declared activities and transient whims of millions of people. Within this unruly pile of data, there may be clues that could help public health researchers track opioid users and issues affecting them.
Abeed Sarker from Emory’s Department of Biomedical Informatics recently published a paper in JAMA Network Open on his analysis of Twitter posts about opioid use. The paper was featured in Popular Science.
Sarker and colleagues from Penn trained a machine learning algorithm on a subset of posts about opioid use, so that the algorithm could analyze a larger body of tweets. They found that Twitter posts about opioids in Pennsylvania, classified by the algorithm, matched the rates of overdose deaths and rates of opioid use measured through national surveys. Their aim is to be able to spot patterns of overdoses faster than prescription drug monitoring programs.
“The findings suggest that automatic processing of social media data, combined with geospatial and temporal information, may provide close to real-time insights into the status and trajectory of the opioid epidemic,” the authors write.
The tweets Sarker’s team looked at were from 2012 to 2015, but he plans to move up the time frame so that he is “not only predicting the past, but the future.” First, he needed to validate his approach, showing that he can filter out unrelated chatter and focus on personal experiences of Tweeters, he says.
“We think this model is more robust than past models because it is more resistant to unrelated chatter — for example, if a celebrity dies from an opioid overdose, there is a lot of social media chatter about it, but that does not mean there is an increase in opioid usage at the population level,” Sarker told Popular Science.
Sarker, funded by a R01 from National Institute on Drug Abuse, says he also wants to analyze other states such as Georgia and other drugs such as stimulants.