On August 23, 2011, some people in New York knew an earthquake was coming before it happened. They weren’t psychic (as far as I know), but digital tweets from their friends in Washington, DC arrived 30 seconds before any seismic rumbles began (1, 2). Afterwards, the U.S. Geological Survey asked people to “tweet if you felt it.” Over 122,000 people responded, providing a detailed map of activity within hours (3). Though phones were dead near the epicenter of the quake, texts kept moving.
Welcome to SOLOMO (SOcial, LOcal, MObile) communication, connecting us instantly through handheld devices. News now literally travels at the speed of light, with words strapped to the backs of zippy electrons. Emergency preparedness and disaster response teams are taking note, using social media to both get and spread the word. The Red Cross has dedicated teams who monitor Facebook and Twitter (4).
While the speed of social–media communication is impressive, its volume is daunting and its content overwhelmingly messy. Besides 300 billion emails (5), each day across the globe we send 200 million tweets(6); search the Web more than four billion times(7); and add 5,000 new blog sites to the 170 million that already exist (8). Ten million people, including the president, belong to FourSquare (9), which delivers personalized offers and local news interactively based on where you are (GPS) and what is nearby. In the 3.5 hours we spend each day “connected” (10) we buy, sell, chat, gossip, work, cheer, complain, and advise. We plan everything from dinner parties to Mideast revolutions, we ask about everything from movie ratings to interplanetary travel, and we monitor progress of local teams, hurricanes, and political races.
In this massive torrent of words, symbols, numbers, jargon and—frankly—noise, are there patterns of information that help us? Is there any hope of sorting through it all?
Actually, yes. SOLOMO media may provide our most effective epidemiological alert system. And the methods being developed to compile and decipher social media may identify what works in medicine faster and more efficiently than through traditional science.
Pulling meaningful droplets from the fire hose
While the science of SOLOMO data mining has advanced quietly in recent years, it may only be scratching the surface of what is possible. Experts refer to the public flow of raw data as a fire hose, from which they are learning how to filter and extract more-manageable streams of content based on what, where, or how often it occurs.
The basics are pretty easy. “Trending” sites show us volumes and frequencies of popular topics over time. Google shows us what people are searching for, and how often. Twitter allows us to see what is being said about a topic (word or phrase). Some sites even let us slice popular topics by locations or types of people (e.g., what are people in New York talking about?).
As mining techniques become more sophisticated, so do the applications they support. Professors at Indiana University report that their algorithm that tracks consumer sentiment is 87% accurate at predicting a company’s stock price (11). Political advisors say small changes in sentiment give them a more accurate real-world indicator of preference than traditional polls (12). Digital marketing experts claim they can not only monitor brand sentiment but also predict human intentions such as intent-to-purchase a specific product or service (13). What consumers say right now may predict success in business and politics tomorrow.
So, how does SOLOMO apply to health and health care?
We see positive glimmers of what the fire hose may bring to our own industry. Epidemiologists now track outbreaks of flu and food-borne illnesses using Tweets (14). Recorded puffs on asthma inhalers (equipped with GPS) now provide useful indications about air quality (15). First Life gathers real-life side-effects or complications from medications reported informally by millions of patients on tweets, blogs, and virtual community sites, revealing far more diversity and detail than any formal clinical trials.
As computers get better and better at NLP (natural language processing), scientists will extract patterns of words that reveal what patients and doctors really experience on a daily basis. Like sentiment about a product, we will learn about functionality and health status in our natural environment. Across hundreds of millions of daily experiences we can learn what makes a difference – for recovery, survival, relapse or quality of life.
Plus, scientists can apply language algorithms to electronic medical records to build best-practice guidelines for use of medications. By examining patterns of outcomes across thousands of patients with almost infinite combinations of characteristics, symptoms and treatments, a team from York University in Toronto is distilling “what happens in real life” into evidence-based recommendations (16). Never before have we had the data, computing power, and methods to observe ourselves in our natural habitat and extract patterns that predict good (or bad) results. Imagine using real-life experience to establish medical guidelines in place of published, controlled studies that today take years!
The impact of social media on health care is a question of when, not if.
Social media will revolutionize how we experience personal and public events. It won’t simply accelerate our receipt of news, but will be inextricably part of a story – changing it by observing and sharing it. Just as Facebook both incited and reported unrest in Egypt, so will these communication streams change our experience of health events and our expectations of health care. Consumers will have more information, on demand, in useable formats, where and when we need it. Consumers themselves may also create some of the collective intelligence that informs each others’ decisions.
We may not know exactly how, but it is certain that SOLOMO will alter every industry. Just imagine if Paul Revere had a twitter account instead of a horse; everyone in the colonies would have seen up-loaded photos of Red Coats rowing ashore. Minute men would have been millisecond men, armed with I-phone apps showing GPS maps of where troops were advancing. Real-time, crowd-shared information changes the game.
At Altarum, our Center for Consumer Choice in Health Care is supporting a collaboration to develop real-time social media indicators of consumerism attitudes and activities in health and health care. Stay tuned for upcoming developments.
- Allain, R. Tweet Waves vs. Seismic Waves. August 26, 2011; (accessed August 31, 2011).
- Fischer, E. Animated map of earthquake Tweets, August 23, 2011. August 25, 2011; (accessed August 31, 2011).
- Working, R. When earthquakes hit, Geological Survey turns to Twitter, Facebook. August 25, 2011; (accessed August 31, 2011).
- US East Coast earthquake generated more Tweets than Osama bin Laden death. August 25, 2011; (accessed August 31, 2011).
- Tschabitscher, H. How Many Emails Are Sent Every Day? (accessed August 31, 2011).
- Your world, more connected. August 11, 2011; (accessed August 31, 2011).
- comScore Reports Global Search Market Growth of 46 Percent in 2009. January 22, 2010; (accessed August 31, 2011).
- Blog Pulse [Web Page]. 2011 Aug 29; Accessed 2011 Aug 31. Available at: http://www.blogpulse.com/.
- Tsotsis, A. Foursquare Now Officially at 10 Million Users. June 20, 2011; (accessed August 31, 2011).
- Schonfeld, E. People Spend As Much Time On Mobile As Reading Newspapers And Magazines. December 15, 2010; (accessed August 31, 2011).
- Ingram, M. Can Twitter Help You Predict the Stock Market? April 6, 2011; (accessed August 31, 2011).
- Brustein, J. Nation’s Political Pulse, Taken Using Net Chatter. October 31, 2010; http://www.nytimes.com/2010/11/01/technology/01sentiment.html (accessed August 31, 2011).
- Swallow, E. Startup Uses Tweets to Predict Purchases. January 25, 2011; (accessed August 31, 2011).
- Twitter Used to Predict Flu Outbreaks. September 29, 2010; (accessed August 31, 2011).
- High-tech asthma inhaler tracks data via GPS. April 19, 2011; (accessed September 01, 2011).
- Cercone, N., An, X., Li, J., and Gu, Z.Finding best evidence for evidence-based best practice recommendations in health care. May, 2008; (accessed September 01, 2011).