Going viral

Machine learning might help us predict where the coronavirus will spread to next

Kinsa is trying to track the spread of the coronavirus using its smart thermometers and machine learning.

Shutterstock

The coronavirus, officially known as COVID-19, is spreading throughout the country, and as of this moment, there are over 560 confirmed cases in 34 U.S. states, there have been 22 deaths. Cruise ships are being quarantined. Airports are seeing far fewer travelers than usual. People are justifiably concerned!

Can a smart thermometer company called Kinsa help ease our fears?

Kinsa has been operating for seven years, and it sells smart thermometers that connect to your phone through an app. If you're worried you or your child is getting sick, you simply take a temperature like you normally would. Once you do that, you'll get a push notification from the app asking for more information about what's going on like what symptoms are present, the severity of these symptoms and more. The app will provide you with the info you need regarding what to do next and will check in with you throughout the following days.

The company tracks the spread of the flu down to the local level, and it's has a program in around 1,500 K-12 schools where it distributes its thermometers. The program is free to the school. You can also find these thermometers on Amazon, at CVS, at Walmart and elsewhere.

Inder Singh, the founder of Kinsa, tells Inverse that his company gets more data on the spread of the flu than anybody else.

"Our mission is to curb the spread of infectious illness. We do that through real-time detection and real-time response," Singh says. "Our work to date has been on flu in the United States, so we've been trying to predict where and when flu will occur. We currently have a really accurate three-week forecast down to a local level."

The average family with two kids has a virus in the household around 29 weeks out of the year. Comparatively, a childless home has a virus in the household around 3 to 4 weeks per year. That's one of the reasons Kinsa focuses heavily on getting its products to people through schools — because families are more likely to come in contact with these types of viruses.

Singh claims Kinsa can predict the spread of the flu three weeks out, and data show its predictions have been consistently accurate. It's expanding that prediction timeline to 12 weeks. Its thermometers are in 1 million households, and it uses machine learning to predict the spread of the flu based on historical data and data its been fed over the years.

Kinsa

What about coronavirus? Singh says he wants to help the Centers for Disease Control and Prevention track the spread of the coronavirus. One of the criticisms of the Trump administration's response to the spread of the coronavirus has been that there hasn't been nearly enough testing happening.

Kinsa could conceivably help us get a better idea of where it's popping up and where it's likely to go next. Singh says the coronavirus and the flu are different, but they may be able to be tracked in the same way.

"The symptomology is different. It's not always fever-inducing, but the methods of analyzing an acute respiratory infection spread—the principles are the same," Singh says. If we can look at the COVID symptomology and apply that to our models, we ought to be able to predict its spread."

Singh says the CDC doesn't have the best data, and it also doesn't have data that goes down to the local level as Kinsa does. The CDC has multi-state regional data on where viruses have spread to and some state-level data that isn't always reliable, he says.

"One of the challenges the CDC has is the timeliness of their influenza data. It's delayed. We get our data in real-time. That alone is a 10-day difference," Singh says. "The second is that the granularity of our data is actually higher."

Kinsa can track how a virus is spreading in the household, a school and a community. Singh says school nurses at the schools they're operating in often inform them that flu cases have dropped significantly since they started working with Kinsa.

"This is not rocket science. If you want to know where and when outbreaks are starting, you need real-time data from people who are getting ill," Singh says. "If you want to know how fast it's spreading, you've got to know how fast it's spreading in a household and in a school. Those are nodes of disease transmission."

As we've previously reported, another problem that's been prevalent since the coronavirus started spreading in the United States has been the proliferation of misinformation about it. Singh says Kinsa can help get people accurate information through its app so there's less confusion about what's going on. Hopefully Singh will be able to give the government the help it needs.

The Inverse analysis

We've come a long way in how we respond to pandemics over the past century or so, but it really seems we haven't adapted as quickly as we could. Why aren't we utilizing technologies like machine learning more if doing so could save lives? We have the technology, we have the experts but so often our response to public health emergencies seem rather low-tech. Companies like Kinsa are showing technology can give us more information on how illnesses are spreading than we've ever been able to get before, and it's time to start taking advantage of that.

Related Tags
Share: