A Canary in a Coal Mine?

Life for a canary in a coal mine could be described in three words - short but useful.

Early coal mines did not feature modern ventilation systems, so miners would bring a caged canary into new coal tunnels. Canaries are very sensitive to methane and carbon monoxide, making them ideal for detecting dangerous gases.

As long as the canary kept singing, the miners knew their air supply was breathable. A dead canary signalled an immediate evacuation.

Recently, I read an article by David Handelsman in Applied Clinical Trials, titled, "A Digital Canary in a Coal Mine?", October 2009 and found the implications fascinating. You can read it (here) but I'm going to try to summarize it for you.

Mr. Handelsman's premise is that as people grow more comfortable with blogs, Facebook, and Twitter (called social media), public comment regarding medical situations will get both more readily available and more inaccurate.

His initial thoughts focused on "text mining" social media to identify very early potential health warning signals. He notes that the speed with which news of the H1N1 influenza virus (swine flu) spread had more to do with social media than for any other reason.

As recently as just one or two years ago, news of the virus would have appeared first on mainstream news sites, followed by emails to friends and family referencing the news. Today, friend networks, blog feeds, tweets, and other messages spread news of the potential flu pandemic via its own "cyber" pandemic.

He claims that there is useful content within social media. For example, the HPV vaccine was approved for use in the spring of 2006. Shortly thereafter, the recipients of the vaccine (primarily teenagers) began documenting their experiences online through various, often personal, social media outlets.

That information, because it was delivered through social media outlets, was available virtually immediately to not only individual readers but also computerized systems that can gather, interpret and analyze this content. Big brother may well be watching you right this very minute!

Today, Drug safety is dependent on an Food and Drug Administration (FDA)-regulated spontaneous adverse event reporting procedure. Various medical clinics, hospitals, doctor's offices gather information and enter it into databases on their own computer systems where eventually it reaches the FDA. None of these various databases are available in real time and furthermore systems such as these are privately-held and subject to privacy regulations.

There is really no doubt that these data sources are extensive and provide the most accurate and comprehensive data to identify true safety signals, but they are very slow in signalling an alarm. The true early warning system lies in the ability to mine and analyze the real-time data available through social media.

To see how this might work, let's look into a potentially very real future, where a new, virulent, flu virus is spreading rapidly, and a new vaccine has been distributed.

It is easy to imagine clinics administering vaccinations around the clock, but how can the success or failure of those efforts be rapidly determined?

Certainly, government officials will look to selected caregivers to report flu-related clinic visits, but even that information will take time to aggregate and review.

Instead, imagine Twitter being used to give an indication about the number of flu patients seen and vaccinated. That information, however inaccurate, is available within minutes. Thus it might be the early warning system that indicates success or failure, and enables health officials to quickly take the most appropriate next steps.

Social media is rapidly evolving, and it is unclear exactly what role it will, or should, play regarding health care and many other, yet unthought about, uses.

What is clear is that social media will play a role in the future. We might embrace it as a digital canary in a coal mine, alerting us to possible problems long before we can detect them ourselves.

How's that for food for thought - you Facebook and Tweeter types?