A New Knowledge-Pushed Mannequin Reveals That Carrying Masks Saves Lives – and the Earlier You Begin, the Higher
Biplav Srivastava, University of South Carolina
Dr. Biplav Srivastava, Professor of Computer Science at the University of South Carolina, and his team have developed a data-driven tool that can be used to demonstrate the effect of wearing masks on COVID-19 cases and deaths. His model uses a variety of data sources to create alternative scenarios that could tell us, “What could have happened?” If a county in the US had a higher or lower rate of mask sticking. In this interview he explains how the model works. its limits and what conclusions we can draw from them.
What does this computer model do?
This is a nationwide tool that can show the effect wearing masks can have. If it’s a county where people regularly wear masks, it will show how many COVID-19 cases and deaths they have avoided. Selecting a county that does not wear masks will show you how many cases and deaths could have been prevented there.
How does it do it
For this we need a lot of data. The New York Times surveyed almost every district in the United States this summer, assigning each of them a mask wearing value from 0 to 5. So this is the heart of the model. We also use New York Times and Johns Hopkins data for real-time case numbers. Census data for demographic data like population size, median age and more; and geographic data to measure distance between counties.
It is based on a mathematical technique called robust synthetic control, which is widely used in drug research where there is a control group and a treatment group.
Take a look at Wyandotte County, Kansas, for example. It has a relatively high mask wear value of around 3.4. Since the model is designed to give us the “what if?” – Scenario reports, it is investigated what would have happened if the score for wearing masks had been reduced to 3.0. This is our “low mask wear” limit, but the user can experiment with other values to see what happens. We came to 3.0 based on analysis of national mask wearing habits. Actual scores ranged from 1.4 to 3.85 with a national average of 2.98.
We can set a date when the mask wearing score will change to 3.0. If we were to set it to run from June 1 to October 1, it would mean Wyandotte County would have had 101.5% more cases and 150 more deaths over that period. It tells the user how many deaths have occurred or been prevented based on a mortality parameter set by the user. In this example a value of 2% has been set.
How does the model create the “what if?” Scenario if it didn’t actually happen? It does this by looking at other counties in the area that have similar demographics and case numbers, but have a lower threshold for wearing masks, come up with a weighted average to create a synthetic control group similar to our county (treatment group) of interest is. The model then examines how much the two groups differed in terms of the number of cases. The two groups are converted into a difference in deaths using the mortality parameter.
What does this tell us about the implications of mask wearing guidelines?
It can be helpful to keep masks on or to implement a mask policy at all times. However, the effects will be greatest if you do it early. If you run this model multiple times with different dates, the effects will be less if you delay the implementation of a mask-wearing policy. So if a county had implemented a mask policy on June 1st, it would have prevented many cases. If it acted on July 1st, it would have less of an impact. If it had acted in August it would still have prevented cases, but a very small number.
What are the limitations of this model?
This tool works better in some countries than others. In general, it works best with counties that are closer to the average as there are closer matches to compare. There’s also a caveat in the sense that the New York Times Mask Compliance poll was conducted over the summer and things are always changing. When other researchers use this tool, they need to consider the changes.
[Deep knowledge, daily. Sign up for The Conversation’s newsletter.]
However, you can see that the implementation of a mask guideline or the regular wearing of masks by the population has a positive effect. And the sooner you do it, the more effective it is.
I would like to acknowledge the work of my team, Sparsh Johri, Kartikaya Srivastava, Chinmayi Appajigowda and Lokesh Johri, in developing this program.
Biplav Srivastava, Professor of Computer Science, University of South Carolina
This article is republished by The Conversation under a Creative Commons license. Read the original article.