Researchers have developed an algorithm capable of mining Twitter hashtags in order to provide early indications of positive and detrimental interactions between medicines.

Twitter: New Way of Spotting Drug Effects

We have seen how Twitter can be used to help resolve conflicts or to ensure the success of a marketing campaign (in French) and how tweets have even been touted as a new literary genre (in French), but it now appears that the blue birdie platform could also help to provide an early warning on drug interactions and also quickly spot any hitherto unknown or potentially dangerous side-effects of medicines. By mining tweets, this valuable information can apparently be made available before it shows up in medical and biological databases such as PubMed.

Twitter has now revealed its potential in the medical field in helping to quickly identify any potentially harmful side-effects of drugs and/or dangerous interactions between medicines

A team of scientists at the University of Vermont in the United States has developed an algorithm they call HashPairMiner, which is able to sift information in tweets and hashtags in order to detect interactions between medicines. ‟"We may not know what exactly the interaction is, but with this approach we can quickly find clear evidence of drugs that are linked together via hashtags," explained research group leader Ahmed Abdeen Hamed in an online statement.

The software programme operates like a search engine: you can enter the names of medicines whose connections and links you wish to investigate.  And the potential applications are numerous. First and foremost, if the programme spots any potentially dangerous interactions between medicines, a general alert can be sounded in a much shorter period than the time it usually takes to send out information to the general public in such cases.

By mining the data in hashtags such as #overprescribed and #skinswelling, the University of Vermont algorithm can detect positive and negative drug interactions.

These days medical publications in digital libraries “suffer infrequent tagging”, points out Hamed. The reason is clear: updating keywords and metadata associated with studies is a laborious manual task and so often neglected. But now, with the Vermont algorithm, the alarm could be sounded very rapidly, helped by Twitter’s viral nature.

Basically, when either a harmful or therapeutic correlation is discovered, this should be used to feed into, and suggest new directions for, ongoing scientific research. “If we're able to detect concerns – say chat about headaches or drops in blood pressure or whatever – that may lead pharmacists or researchers to a hypothesis that can be followed up by a clinical trial or other medical test,” explains Ahmed Abdeen Hamed. The data scientist cites one surprising discovery the team made using the hashtag mining approach. Delving into the hashtag #Alzheimer, the algorithm revealed a correlation between ibuprofen, medical cannabis and treatment for Alzheimer’s disease. When combined, the therapeutic substances contained in ibuprofen and cannabis are apparently able to slow down the brain degeneration caused by Alzheimer’s, while the ibuprofen helps to neutralise some unwanted side-effects of cannabis use. This result “appeared on Twitter before PubMed!” enthused Hamed.

By Pauline Canteneur