How Can Twitter Can Be Used To Monitor Drug Use And HIV?
Social media has revolutionized the way we communicate, but can it be used to pick out at-risk areas for drug abuse, risky behaviors and HIV? According to a new study, the answer may well be yes. There are limitations to the approach, but our penchant for broadcasting our thoughts, intended actions and feelings through publicly accessible sites like Twitter could provide a pre-emptive warning about areas particularly at risk for a wide range of health issues. In the past, the approach has been suggested as a method of predicting influenza outbreaks, but the new study suggests that a spike in the number of people using key phrases like “get high” or “sex” in a particular area may be used to determine where to target public health programs relating to drug use and HIV.
The study comes from a new center at UCLA intended as a multi-disciplinary effort to investigate the potential of social media and cell phones to enable prediction (and hopefully the changing) of behavior in the population. The new piece of research is primarily focused on the prevention of HIV, but one of the key ideas the research was based on is the link between risk behaviors like sex or drug use and how they spread. The methodology of the study was fairly straightforward: collect a lot of tweets, use an algorithm to search for words and phrases indicating risky behavior and drug use (with a focus primarily on stimulant use) and plot them on a map. The finalized map was then compared to HIV data to see if the prime areas of suggestive tweeting correlated with the prevalence of HIV. In total, the researchers collected over 550 million tweets from May to December 2012, and compared the final map to an interactive map of HIV prevalence in the US. However, the most up-to-date information available when the researchers were conducting the study was from 2009.
Does Twitter Work For Tracking?
The sample they analyzed contained over 8,500 tweets relating to sexual behavior and over 1,300 suggesting stimulant use. By state, California, Texas, New York and Florida had the biggest proportion of suggestive tweets, but the main hotspots identified in terms of the raw number of tweets identified were the District of Columbia, Delaware, Louisiana and South Carolina. The largest per-capita rates (taking state populations into account) were found in Utah, North Dakota and Nevada. However, for the purposes of the analysis, the focus was on the location of tweets in terms of counties, since state-level data wouldn’t offer much potential for a truly targeted approach.
The researchers then compared their findings to the data on HIV prevalence, and found a statistically significant relationship between the locations with the most risk-related tweets and reported cases of HIV. Although this is very encouraging, a big issue with the study is that the HIV data came from 2009, three years before the tweets were collected. While it stands to reason that the areas with the largest numbers of risk-related tweets would be those with the greatest prevalence of HIV, the problem of old data does hold back the implications of this research. If the prevalence of HIV didn’t change much over the intervening three years, then the study shows a correlation, but it doesn’t necessarily mean it’s possible to predict an outbreak.
Social Media Strategy In The Future
The biggest problem with the approach from the new study is getting up-to-date information about HIV prevalence, or indeed that of any condition or issue being monitored. The researchers suggest that for the approach to work there would need to be some form of “gold standard” for regularly updated information, and this is well within the realm of possibility. If a similar study conducted with up-to-date data comes up with comparable findings, there will undoubtedly be more incentive to invest money in the approach.
The other issue with the potential for applying this approach on a larger scale is the sheer volume of data required. The research used over half a billion tweets for a period of six months, and each one had to be checked using the algorithm to produce the final data. It’s called “big data” for obvious reasons, but it’s becoming increasingly important because we can theoretically collect and use this much information for the purposes of monitoring potential disease outbreaks. Studies like this show that it’s possible, but whether we’re really ready to start monitoring huge volumes of data like this all the time and for myriad purposes is another matter entirely.
Prospective Prevention Of Drug Addiction
If we set aside the organizational and logistical challenges presented by this approach—since it will be feasible in the near future—it could have profound consequences for how we target treatment, education and prevention programs for drug addiction. It’s been suggested as a method to help prevent the use of Adderall as a “study drug,” and if applied as a more wide-ranging strategy, it could help us identify areas where drug use is burgeoning. Effectively, it presents the very real possibility for the widespread prevention of drug addiction, not just the treatment of it.
Read How To Determine If you Have A Social Media Addiction