Graham Dodge (Illustration by Lennart Andresen)

The spread of Ebola continues to garner attention after the first patient diagnosed on American soil died from the disease Wednesday. But one local start-up sees an opportunity to monitor outbreaks — from allergies and stomach bugs, to Ebola and the Enterovirus — using social media. Sickweather is capitalizing on the era of oversharing by tracking posts about health symptoms, such as a fever or cough, on Facebook and Twitter. The company then plots them on a map to monitor where an illness seems to be spreading.

The Baltimore-based firm, which has just a handful of employees, raised $350,000 from a group of angel investors last month. It’s the brainchild of co-founder and chief executive Graham Dodge, who previously worked in design and marketing for such outfits as Crime.com, MTV Networks and accounting firm KatzAbosch. He discussed Sickweather’s technology and its potential implications for public health officials with Capital Business.

What is the premise of Sickweather?

Sickweather is the Doppler Radar for sickness. That’s the simplest explanation of the premise. Basically what that means is we’re tracking illness in real time and we’re using data that we’re distilling from social media to do that. It’s really the first time in the history of mankind we’re able to do that thanks to people sharing their symptoms on social media.

A screen capture of the Sickweather app shows illnesses that are present in a users' location. (Courtesy of Sickweather)

How does it work technically?

We have a patent-pending process that searches the Twitter and Facebook application programming interfaces and that gives us access to all the public data that’s available. So we’re not accessing information that’s behind privacy settings.

Our service says give me all the tweets related to the flu, and then we distill that down to all people talking about having the flu. Then we’re able with geogtag information to plot it on a map in real time to create these predictive models for where illnesses are traveling.

So you can not only tell where an illness is now, but where it is headed?

We use some modeling that’s based on historical data that we have from our own data as well as data we collect from the Centers for Disease Control to tell what the trends in an area were previously. If we see them trending differently this week, for example, we can model out how that might play out.

We see it often in the spring and summer with stomach viruses. We see it start, for example, in D.C. and then move up to Baltimore or down to Richmond. That’s very common to see in our map. The vector of that is not necessarily the same person reporting and moving about. It’s most likely as people become infected they are talking about it, so we’re mapping that social network of people contracting that illness as it spreads.

How can you be sure the information on Twitter or Facebook is accurate?

That inaccuracy really fades away when you look at the data in aggregate. While there may be a few data points here and there of what we call false positives, where someone has misdiagnosed themselves, in aggregate the trends do emerge accurately.

Where did this idea come from?

Basically I was sick with a stomach virus about three years ago, and I wanted to know if something was going around or I had food poisoning. There was no local data from the CDC or my community health department that showed me, in real time, if there was a stomach virus going around where I lived. So as someone who previously mapped crime data. . . it occurred to me that social media could be used as a source of information for that data.

Has all the talk about Ebola shown up on your radar?

No it hasn’t. We aren’t actively searching Ebola right now. We are looking at the data for any kind of signal we could use for tracking it, but despite all of the noise about it, it is not as widespread as the flu or even Enterovirus. So it would be a very weak signal to try to track in the U.S. If it does spread in the U.S., we will be tracking it and trying to provide early warning for it.

If there were an outbreak, how would Sickweather track that?

Initially it would probably show up via the symptoms of Ebola, such as high fever, body aches, things like that, which unfortunately are very similar to flu symptoms. So Ebola would show up as spikes in those related symptoms and we would start to overlay [the reports of symptoms with] what we know about official cases. We would then use that to determine where Ebola may be emerging next.

You mention the Enterovirus, which has impacted many children across the country. Is Sickweather tracking that?

Right now we’re tracking the symptoms of it and we’re seeing how people are being diagnosed with Enterovirus. We are doing an offline study right now that would show how those [symptoms and diagnoses] are all related, and retool our system to see how we can track for Enterovirus more specifically. That will happen much sooner than we would be tracking Ebola.

Have you worked with public health professionals on this?

Something that the CDC does have an issue understanding is the granularity of tracking illness at the community level in real time. We’ve been talking with the CDC, in fact, about how we can help in that surveillance. They have a project called Operation Dragon Fire that could be the mechanism in which we work together.

It takes the CDC about two weeks to collect their data, and then the media picks up on it. That can be the entire viral lifecycle, two weeks, so at that point the disease has already spread and everything is more reactionary. In 2012, we were able to detect flu season six weeks before the CDC picked up on it. It was that success that gave us some attention early on.

Is the flu coming to an area near you? Health start-up Sickweather uses Twitter and Facebook posts about symptoms to determine when and where an illness is spreading. Chief executive Graham Dodge said it could help track emerging threats like Ebola and Enterovirus.