A lot has happened in a week as the state response to the coronavirus outbreak.
On March 15 Governor Doug Ducey and State School Superintendent Kathy Hoffman announced that all Arizona schools would close through March 27. Five days later they extended the closure until April 10, and there are indications that schools are preparing to close for the remainder of the school year.
Ducey has also shuttered bars, movie theaters, and gyms in counties that have at least 1 positive case of COVID-19. The same order restricted restaurants to take-out and delivery only in the affected counties.
The effects of the COVID-19 can be seen everywhere, except for the number of positive cases reported by the state… 508 as of March 26.
Even with what appears to be a very low infection rate, state authorities anticipate needing an additional 13,000 hospital beds and 1,500 more ICU beds to properly handle the crisis. In the same press conference, it was projected that Arizona would see its peak infection rates in mid-April and peak hospitalization rates in May.
So how does the state come up with these numbers? Computer modeling.
A simple definition of a computer model is an application that takes in data and makes predictions from it. They are used every day, driving the decisions of businesses and governments worldwide. The most well-known computer models in the world power weather forecasting.
Al Katawazi is a data scientist and developer that has worked on projects with the state and major tech firms like Amazon. He has developed his own COVID-19 propagation model for Arizona that is in line with the state’s call for more hospital beds. His model tracks the virus’ spread across the state in what he calls a worst-case scenario, one in which it spreads uninhibited by social distancing or the closure of social spaces.
“In this scenario, ICU beds would reach saturation about 12 days from now.” He said.
One of the most important numbers that the model takes in is the R0, (pronounced R-Naught). It’s the average number of people that will catch an infectious disease from an already infected person. For COVID-19, that number is 2.3 according to experts. For comparison, the common flu’s R0 is about 1.3.
While the difference does not seem like much, over time, COVID-19’s infection rate rises exponentially faster than the flu, to the point where even if only 1% of infected people require an ICU bed, healthcare facilities would still be saturated.
Katawazi recognizes the limitations of modeling. He recognizes that models are only as good as the data that goes into them, and when dealing with human beings, there are always more variables that cannot be tracked than can. However, as in the case of COVID-19, they are very good at giving a prediction that can help state health authorities and medical providers from keeping the system from being overwhelmed.
