It takes time for general knowledge of an increase in opioid-related deaths to translate into official CDC statistics. Moreover, it takes time for health organizations to gather and analyze data, and that has an effect on how quickly the US can tackle the opioid crisis.
However, that does not mean you have to sit and wait for official CDC reports to begin tackling the problem. Artificial intelligence (AI) is allowing addiction researchers to gather information more quickly by analyzing data that is by nature fast and up-to-the-minute, namely, social media data. It is possible these AI tools will help direct addiction treatment resources where they can do the most good.
Applying AI to Twitter to Analyze Opioid Use Trends
About half a billion messages are posted to Twitter every day, and many tweets are tagged based on the user’s location. It is possible that Twitter messages could offer an early warning system of sorts that could prompt smaller, localized public health initiatives. Social media is enormous, and of course, there is plenty of non-drug related information on it. However, AI tools may be able to filter out messages that are not opioid-related and apply quick analysis to relevant posts to offer more detailed information about where and when acute opioid crises develop.
These tools are designed to preserve confidentiality by filtering out individual user information, yet give public health officials more “granular” information about where and when local opioid crises arise.
Keywords, Locations Offer Quick Epidemiological Snapshots
Tweets related to opioid abuse tend to use certain code words, and part of AI’s power comes from human addiction experts “educating” the software on words that are commonly used on social media to refer to opioids. Words like “syrup” tend to be code for codeine, while deadly fentanyl sometimes goes by the term “dummies.” Of course, classic slang terms for opioids like “percs” and “dope” are commonly used.
AI tools being developed by researchers in New York, New Jersey, and Utah include keywords related to opioids that are curated by physicians and toxicologists with “boots on the ground” who learn the terms through their everyday practices. Keywords coupled with geographical tags can potentially identify localized opioid crises far more quickly than traditional tools like CDC surveys can.
How Well It Works
Initial AI studies on social media posts have been compared to geographic distribution data collected through National Surveys on Drug Usage and Health from 2013 through 2015, and the state-by-state correlation between analyzed Twitter data and official data was highly significant, especially for younger drug users (who are likelier to use social media). Correlation between AI results and national survey results was driven by the discussion of the use of opioids and held up even after researchers controlled for geographic variation in Twitter use. The biggest problem with the use of AI tools is the huge amount of irrelevant data that must be sifted through to get to meaningful messages about opioid use, but results are promising nonetheless.
Toward Better Long-Term Outcomes
If huge numbers of tweets can be used to pinpoint the location of opioid crises more quickly than official addiction data, so what? Knowing where problems exist and doing something about it are two different things. However, a partnership between IBM Watson Health and MAP Health Management is taking the data a step further and aims to fill gaps in long-term care.
Data analyzed by Watson and MAP is specifically being packaged to help care managers, insurers, and addiction treatment experts to help prevent relapses and ensure better long-term results for addiction treatment. As the software continues to “learn,” it will become easier to identify people with higher risk of relapse so that local healthcare resources can be prepared to respond more quickly.
Addiction treatment works more effectively the sooner it is begun. You know there is a deadly opioid abuse epidemic, and you know that opioids kill more quickly than official data can be gathered and analyzed. New AI tools may help addiction treatment researchers learn where and how quickly local crises develop so that resources can be directed promptly.
Of course, people with opioid use disorders and their loved ones do not need AI or official data to tell them the problem is real. If you are struggling with opioid misuse, the sooner you reach out for help, the better your chances for long-term recovery. We invite you to learn about our admissions. There is no obligation, and it could put you one step closer to putting opioid abuse in the past.
- Understanding The Difference Between Alcohol Use and Alcoholism - December 12, 2018
- Do You Need Gender-Specific Addiction Treatment? - October 9, 2018
- Is Science Closer to a Cure for Alcohol Use Disorder? - August 30, 2018
- Why Is Alcohol Use Disorder On the Rise Among Women? - December 6, 2017
- Survey Shows 12 Percent of Americans Have an Opioid-Addicted Family Member - December 5, 2017
- Opioid Epidemic Costs Estimated in the Billions - December 4, 2017
- Yale Announces “Innovation to Impact” Program to Spur Substance Abuse Treatment Solutions - December 1, 2017
- Why Prevention Is an Essential Weapon in the Fight against Substance Abuse - November 30, 2017
- What Is the Difference between Drug Addiction and Behavioral Addiction? - November 29, 2017
- Marriage, Addiction, and Divorce: What You Should Know - November 28, 2017