How accurate are models? As accurate as the Tarot?

CNN had a headline this week stating that COVID-19 deaths will rise almost 80% by February. This is based on an influential model – I do not understand why the model from the Institute for Health Metrics and Evaluation at the University of Washington School of Medicine is considered influential but let us assume that it is.

Models are models. https://www.nature.com/articles/s41599-020-00553-4 People crank equations with anticipated events and expectations and voila, a prediction occurs. The United States is currently reporting nearly 218,000 COVID-19 deaths (in reality, probably much closer to 175,000 real deaths from the SARS-CoV-2 virus). Regardless of the real number of coronavirus deaths, is it likely that we will see a near doubling (80% increase) in the next hundred days?

The Influential Model

https://covid19.healthdata.org/united-states-of-america?view=total-deaths&tab=trend is the website from which the influential model determined the prediction of nearly 390,000 coronavirus deaths by February 1st, 2021. This model reflects three scenarios.

One scenario that is modeled is if masks are not mandated. I cannot in my wildest dreams seeing any situation within the next 100 days that all governments (local, state, and federal) would make masks optional.

This model is called mandatory easing. It assumes that mandates for facial masks would be eased and no future requirement to adopt facial masks would be implemented. Does this make sense? This model projects over 475,000 COVID-19 deaths by February 1st.

There is a caveat associated with this model that was slipped into the article about future infections will be with younger people and the latest, revised, predictions of deaths are not under 400,000 coronavirus deaths by February 1st. That is a huge adjustment based on the population that might become infected. Models are models and subject to humans’ ability to predict accurately.

A second scenario is that our current policy of facial masks remains in effect. If by chance, face mask restrictions are lifted for whatever reasons, a mandatory reimposition of face mask-wearing would be instituted whenever the death rate for six weeks exceeds eight/million.

With fifteen weeks to go to approach the end of January, there would not be much time to remove and reimpose a facial mask mandate. I am guessing that the current policy will not ease any mandates for wearing face masks. The model estimates nearly 390,000 COVID-19 deaths – just under the recently updated model projections that a larger number of younger people will become infected which changed the 475,000 projected deaths to 395,000 deaths.

The first question that pops into my mind is the validity of this projection as it applies to younger people becoming infected and not dying from the virus. Should not this prediction of 389,000 deaths also be reduced? No update in this article about that. https://www.cnn.com/2020/10/15/health/us-coronavirus-thursday/index.html

The last of the three scenarios identified in this influential model is for universal mask mandates. Everyone (95%) wears a mask in public. Is that not what we are doing now? I see people wearing masks in every store or facility I enter. Signs on the doors tell potential entrants that face masks must be worn.

Why is this scenario not considered to be the current projection? In this prediction, by February 1st, 314,000 coronavirus deaths can be expected. That is nearly 100,000 deaths more than today. 100 days and 100,000 more deaths.

At one point in time, the United States’ daily death rate was over 1000 daily deaths. The last seven-day moving average in which 1000 died in one day was August 22nd. https://www.worldometers.info/coronavirus/country/us/

For the last 30 days, the United States, on a seven-day moving average, has seen less than 800 deaths daily. What is the event (or events) that will catapult COVID-19 deaths spiraling upward?
Yes, new COVID-19 cases are setting records in a few states, but the death statistics in those states are not rising at the same trend as new cases.

Conclusion

Models are models! They are designed and fed data by humans. It is like our weather prediction models. How many times have they been correct – even seven days in advance, much less 100 days in advance?

I have seen several weather predictions of well over 50% chance of rain fizzle to nothing on the day of the prediction. I expect that there are occasions that we may or may not get rain when the forecast predicts 50% and above. But I should expect at least a 50% chance of seeing rain on one of those days. I have not during 2020. I tracked rain predictions closely this year for watering my yard and adding water to my pool. I was disappointed many times! Oh, it must be the weather prediction models that did not incorporate people staying indoors so they would not know if it rained or not!

Live Longer & Enjoy Life! – Red O’Laughlin – RedOLaughlin.com

 

2 Responses

  1. I see many people not wearing masks. They either don’t know or don’t care. This is exaggerated by the confusion created by the government’s “scientists” saying first they aren’t needed and now saying they are. This is further enhanced by the youngsters believing they will live forever and being told if they do catch it they won’t be real sick.

Leave a Reply

Your email address will not be published. Required fields are marked *