Guest Article: Can Twitter help stop or slow down human virsuses?

Chris Thorman, who blogs about medical software at Software Advice, gave me a call recently and talked to me about Twitter’s potential use in tracking disease outbreaks. As evidenced by Twitter’s front-and-center role in the “Second Iranian Revolution,” the micro-blogging tool has emerged as a serious communication channel. Chris talked to me about how he thinks Twitter is every bit as relevant in medicine as it is in politics, business and personal communications. Specifically, we’re convinced Twitter could play a big role in epidemiology – the branch of medicine that deals with the study of the causes, distribution, and control of disease in populations. I loved his ideas so much that I invited him to write about how this might work and he created the following guest posting. Enjoy.

The combination of Twitter and epidemiology presents an interesting opportunity: What if doctors tweeted about symptoms they observed and diagnoses they made? What if that information was aggregated in a way that helped track disease outbreaks in real-time, share treatment plans, and save lives?

Combining health data with social media tools to track disease outbreaks is a simple concept. Executing this in the real world, however, is extremely difficult. Dr. John Snow, the father of epidemiology, tracked the 1854 Broad Street Cholera outbreak in London’s SoHo district. Dr. Snow recorded the locations of the 600 cases of Cholera on a spot map, spoke with SoHo residents to learn more about how the disease spread, and finally concluded based on the locations of the cases and personal encounters that the Broad Street water pump was the main agent responsible for spreading Cholera.

Fast forward to 2009. How can we combine social media tools with electronic medical records (EMRs) to help track disease outbreaks like the Swine flu?

Getting Reliable Data

Today, Dr. Snow’s interaction with SoHo residents could theoretically have been done via his Twitter feed. The modern day equivalent of Dr. Snow’s spot map may look something like this Google Map/Twitter “mash up”, which combines the visual affect of a digital map with the real time tweets from Twitter users talking about the Swine flu. However, when looking at that mash up, you’ll see that there is no filtering of the tweets’ relevance to an epidemiologist’s needs. Any communication referencing Swine flu, from jokes to local news stories, appears on the map.

Bio-surveillance company Veratect is trying to track diseases in a similar way by monitoring social media traffic on blogs and Twitter feeds talking about the Swine flu. Even though Veratect’s project is much more robust than the Swine flu map mash up, it still suffers from a high signal-to-noise ratio.

What exactly constitutes evidence of a disease outbreak when you’re monitoring social media traffic? How can doctors and epidemiologists verify the information being sent in?

@YourDoctor

Imagine this. Doctors around the world are conducting their rounds and examining patients on electronic medical records, which document diagnosis codes. As the physician makes a diagnosis or documents symptoms, he has the option to “tweet” that observation. This allows other doctors to follow that feed and observe trends. Even better, epidemiology-specific analytics can be layered on top of the feeds to recognize patterns as they develop.

Here is a fictionalized example of how analysis of a disease outbreak might unfold if doctors adopted Twitter as a method of communication:

This adoption by doctors would need include a verification system that only allows trusted or authenticated users to tweet about information contained in the EMRs. What we’re trying to avoid is aggregating a whole mess of data related to a particular disease. Authenticating users to make sure they are who they say they are avoids this problem.

With a uniform set of diagnosis codes and a proper authentication system, suddenly the trending data sent out by these verified doctors’ tweets goes from speculative to extremely reliable.

Unique diagnosis codes could also be translated into other languages, making worldwide tracking of diseases a reality. Personal communication between doctors and epidemiologists would still be hampered by language barriers but at least every user of this system would have access to the same reliable information in their native language.

Twitter users employ “hash tags” to help group their tweets together. By using the # symbol before a word, that word becomes a hash tag and links each tweet of said tag together. Twitter groups these hash tags together as trending topics, allowing anyone to click on a hash tag and bring up every tweet that references it. Epidemiologists could aggregate disease data coming from doctors in a similar way, substituting the Twitter hash tag search for a diagnosis code search.

The Advantage of Real-Time Information

During any disease outbreak, time is of the essence. Many government and health agencies around the world aggregate their data on potential outbreaks but do so on a weekly or semi-weekly basis at best. The technologically primitive nature of the vast majority of the world’s health care systems prohibits catching most outbreaks in their infancy. Even if a disease outbreak is discovered, that outbreak may only be realized at the local or regional level. When you’re talking about potentially killer diseases – Swine flu as a recent example – an advance warning of even a couple of days could mean thousands of lives would be saved.

The real time nature of a Twitter EMR system would allow epidemiologists to get a jump on disease outbreaks. Much like the trending topics section of Twitter, symptoms and diagnoses could be tracked by their frequency as they’re submitted by doctors. Algorithms can be developed to push relevant diseases and their diagnosis codes to the top of epidemiologists’ tracking lists.

Limiting Privacy Concerns

Naturally, there are going to be privacy concerns about doctors tweeting patient information out into the digital world. However, no personal identifying information is required to track diseases in this scenario. The only name associated with the posting of this health information would be the doctor’s. Even that may be an alias.

The combination of social media and EMRs, in some form or another, will undoubtedly be part of the future of tracking disease outbreaks. The how and when of that process remains complicated, dependent on health agencies, governments and the doctors themselves to implement the appropriate systems. However, the “viral” spread of Twitter leads us to believe that physicians may not have to wait around for bureaucracies to organize an epidemiological social network. Like the Iranian opposition party, they may organize it themselves with Twitter.