In the past year, COVID-19 has brought most of the planet to a standstill. In the United States alone, 30 million cases have been reported, and as of February 2021, there have been more than 500,000 deaths.

At last, there is some reason for optimism. There are now three effective vaccines and their production and distribution are increasing, bringing us closer to herd immunity. But there are also new variants of the virus, and although cases dropped sharply in January and February, they now seem to have leveled off.

The fight is far from over.

In February, we announced the launch of ContagionNet, a nonprofit subsidiary of DataRobot. ContagionNet is pioneering a fast, accurate, affordable test for COVID-19, one that will help public health officials better track the path of the virus, and then take steps to stop it.

We started this effort in the middle of 2020 and uncovered two big weaknesses in the COVID-19 data: 

• Missing data for cases, hospitalizations, treatments, and more, which makes it difficult to fully grasp the scale of the pandemic. 
• No way to combine affordable, low-tech antigen tests with a high-tech digital platform that can use data to better track the pandemic.

I recently had the opportunity to speak with Sally Embrey, the VP of Public Health and Medical Technologies at DataRobot, who is leading ContagionNet’s work.

As Sally explains, current COVID-19 detection tests work by amplifying a copy of the virus until it’s detectable. Antigen tests look for antibodies to the virus, evidence that someone has, or has had, COVID-19.

By looking for two specific types of antibodies, ContagionNet goes beyond the yes or no test result and provides  information about viral load. 

“Being a data-driven company, we wanted to get more information out of our tests. By looking for three different antigens—one for the spike protein (named after COVID-19’s distinctive spikes), one for a nucleocapsid protein, and one for the same nucleocapsid protein at half the concentration.

“That variability allows us to understand whether someone has a low, medium, or high viral load. We can then translate that into how contagious a person is.”

Viral load is helpful information because many people who carry COVID-19 and have antigens in their bloodstream show no symptoms. “No contact-tracing effort can ever work with people who are asymptomatic,” Sally said. “Finding out who is actively contagious is a huge hurdle when it comes to identifying who has COVID-19.”

Currently, not nearly enough people are tested. Either the tests are unaffordable (prices range from $20 to $800) or people wait until they have symptoms. Sally’s team is working on a test that has a target price of three dollars.

So where do we stand with this effort? “We’ve spun up a lab in San Diego where people have been developing lateral flow assays (the type of test we’re using) for many years,” Sally said. “That lab can get to production numbers of 100,000 assays a day. That’s not enough in the scheme of things, so we are looking to partner with additional manufacturing organizations.”

For Sally, this work is personal. Her parents have had their retirement plans disrupted by COVID-19 and one of her brothers was furloughed for eight months.

“So many people have been impacted by this,” she said. “One of the scariest things about COVID-19 is that we still don’t know so much. I’ve felt so honored to work on this project for the past six months.”To hear more about understanding the changing face of data in a COVID world, check out Datarobot.com/podcast or http://datarobot.buzzsprout.com/. You can also listen everywhere you already enjoy podcasts, including Apple, Spotify, Stitcher, and Google.