While looking at the virus testing shortage through the lens of software, it seemed like there must be a better solution than brute force testing everyone individually. I realized that with some small tweaks to our testing procedure we may be able to screen vastly more people with the same supply of tests. South Korea's impressive ability to flatten the curve was because of widespread testing so I knew that's where I should focus my efforts.
What it does
In a nutshell, its a simple procedure change of batching swabs together so multiple people can share the same test. If the test is negative, everyone in the batch is cleared simultaneously and many tests are saved. If the test is positive, then just the people in that (hopefully) significantly smaller batch are then tested individually. The algorithm depends on averaging costs and assuming that most people in a test will be negative. If most people in a group are sick you actually use more tests, but on average only a few people in a group will test positive so a huge amount of testing can be saved. People that are obviously sick should still be tested directly. The main application of this method is regular widespread screening to find the asymptomatic carriers like hospital staff, schools, airplanes, offices, etc..
How I built it
Challenges I ran into
Trying to find out how many swabs can potentially be combined and still reliably trigger the viral threshold needed for detection by a test. I could not find an answer so a lab will likely just have to start with a batch of 2 people and work up from there experimentally.
Accomplishments that I'm proud of
The simplicity of the idea. No new technology needs to be developed, it's just a simple procedural change that can be implemented immediately and maybe make a dent in this disaster.
What I learned
What's next for Boost Testing Capacity Tenfold Using Batch Testing
If a lab can confirm batch testing in even as little as a group of 2 is possible, I'll get some coding friends together to start brainstorming the most efficient algorithms asap. Please message me with any ideas or if you want to help!
Log in or sign up for Devpost to join the conversation.