Using Twitter to monitor and measure public health and wellbeing
Social media listening is finding applications in healthcare. It’s a powerful example of modern marketing and public relations.
Social media listening has become a booming business in the past five years. It provides insights for campaigns and an early warning alarm for issues and crises.
A growing third-party tool market exists to monitor conversations in the public area of the web and networks such as Twitter.
A recent Forrester report named Sprinklr, NetBase, and Synthesio as Leaders; Crimson Hexagon, Digimind, Talkwalker, Sysomos, and Brandwatch as Strong Performers; and Linkfluence and Zignal Labs as Contenders.
Tools are limited to the publicly available data on the internet and relationships that the vendor has brokered with technology platforms. There’s an over reliance on Twitter for insight in marketing and public relations because it is easily accessible and in its native form is free.
That said Twitter is a good place to seek out real-time information on an organisation, brand, market or issue but data should be filtered against your audience profile, and cross referenced against other data sets.
You can query Twitter using free native search tools. An advanced version allows variables such as Boolean keyword queries, accounts and location based on geotags and date ranges to be queried. The third-party tool market goes a stage further and enables more sophisticated analysis and data manipulation.
A new research paper published by MDPI called Using Twitter for Public Health Surveillance Monitoring and Prediction to Public Response (opens as a PDF) reviewed 28 separate studies of Twitter monitoring in healthcare.
The use of social media in healthcare has been fraught with regulatory issues however it is starting to find applications such as monitoring disease, public reaction to health issues, and disease outbreak monitoring.
The paper, written by an international multi-disciplinary group, examined a variety of techniques. It also explored the use of machine learning to analyse content using keywords, linguistic analysis and pattern matching.
Disease monitoring sought out ailment related tweets such as influenza and correlated these with the outbreak of locations using geotagged data, or analysis of time zone based on the account’s usage.
A study by researchers at John Hopkins University in Baltimore in 2011 (opens as a PDF) found results consistent with other data such as Google Flu Trends, a service that predicts influenza based on search data.
Another study by John Hopkins University in 2014 (opens as a PDF) used natural language processing to identity and analyse four mental health conditions including post-traumatic stress disorder (PTSD), depression, bipolar disorder, and seasonal affective disorder (SAD). It demonstrated that natural language processing can be used to gain insight into specific disorders as well as undiagnosed conditions.
The reaction to public health issues such as e-cigarettes and the side effects of drugs has been analysed in a variety of ways. These include manual keyword analysis and coding, and sentiment analysis.
There’s strong analysis to suggest that Twitter data can be used to predict the outbreak and spread of disease.
A study of floods in Australia from March 2010 to February 2011 identified how public health information is shared in a crisis. An analysis of Ebola related tweets in 2014 shows that its spread could have been predicted.
There can be little doubt of the power of Twitter as a low cost means to monitor public health however it’s an imperfect science in which methodologies continue to be developed.
If you’re interested in exploring this area further, I recommend Social Monitoring for Public Health by Michael J. Paul and Mark Dredze. They’re the research team behind the University in Baltimore influenza study.
