I haven’t made any COVID-related visualizations this year–for the obvious reason that I’m not a health expert–but I did see a visualization recently that I thought could be improved, so I reached out to the authors and created a new version.

Here’s the quick background.

A relative sent me an email a couple of weeks ago saying that, “If everyone single person wore a mask, even an imperfect one, then this virus will be destroyed. See the contour graph in the article. Why is this not in publicized is beyond comprehension.”

She pointed to this article in Fast.ai that showed with this contour graph of the relationship between the percent of the population who wears a face mask and the effectiveness of that mask. As we move to the darker blue area–more people wearing masks and more effective masks–we can drive the transmission rate of the virus to zero.

My response to my relative’s email was that maybe the conclusion isn’t publicized not because the message is beyond comprehension, but because the math and the jargon in the original image is beyond comprehension. I know a bit of math and I know how to read a contour graph, but I’m no epidemiologist or biologist, so the formulas and jargon are just confusing.

I dug around and found the original paper from a research team in Hong Kong. I grabbed a screenshot of the original image (interestingly, on page 10 of the Supplementary Material), dropped it into PowerPoint, and added explanatory text in and around the graph to match the conclusions in the Fast.ai article and the academic paper.

Before sharing, I sent an email to the Fast.ai authors (Jeremy Howard and Trisha Greenhalgh) and the primary investigator of the paper for their thoughts and reactions. I didn’t want to share an image that was incorrect or misleading. Surprisingly, all three responded almost immediately with praise and some relatively minor comments. Thus a revised version was born. (The folks at Fast.ai even posted my version to the original article.)

If you’re interested in sharing, you can download this zip file with the image in PDF, PNG, and TIFF file formats.

UPDATE, 5/7/2020: Jeremy Howard from Fast.ai wrote today that the “R0” in the original should be replaced with just “R”. I have updated the image and the zip file.

PS. This is the first in my efforts to write more, shorter blog posts. I tend to ramble a bit, so we’ll see how I do.