& # 39; Uncertainty is the only certainty & # 39;


SEATTLE – A statistical model cited by the White House produced a slightly less bleak figure on Monday for a first wave of deaths from the US coronavirus pandemic. A projection designed to help officials plan for the worst, including having enough hospital staff, beds, and fans.

The only problem with this relatively good news? It is almost certainly wrong. All models are wrong. Some are less wrong than others, and those are the ones public health officials trust.

Welcome to the world of bear and grimace modeling.

"The key is that you want to know what's going on in the future," said NASA's chief climate modeler Gavin Schmidt. "In the absence of a time machine, you will have to use a model."

Meteorologists use models. Climate scientists use them. Supermarkets use them.

As leaders try to control the coronavirus outbreak, they turn to numerous mathematical models to help them figure out what might happen, keyword, power, next, and what they should try to do now to contain and prepare for the spread.

The model updated this week by the University of Washington, the one most mentioned by US health officials. USA In White House briefings, he predicts that daily deaths in the US USA They will peak in mid-April and then decline during the summer.

Its latest projection shows that between 49,431 and 136,401 Americans will die in the first wave, which will last until the summer. That's a huge variety of 87,000. But just days before, the same team had a range of nearly 138,000, with 177,866 as the highest number of deaths. Officials accredit social distancing.

The latest calculations are based on better data on how the virus works, more information on how people act, and more cities as examples. For example, new data from Italy and Spain suggest that social distancing is working even better than expected to stop the spread of the virus.

The time it took for the epidemic to peak, that is, for those deaths to begin to decline, was shorter in those Italian and Spanish cities than in Wuhan, China, said Dr. Christopher Murray of the University of Washington, who developed the model. .

So how does modeling work? Take everything we know about how the coronavirus is spreading, when it is deadly and when it is not, when symptoms appear and when they do not.

Then consider everything we know about how people react, social estrangement, orders to stay home, and other soft human factors.

Now add everything we know about testing, treating disease, and equipment shortages. Finally, mix large amounts of uncertainty at all levels.

Squeeze all those thousands of data points into incredibly complex mathematical equations and voila, this is what will happen next with the pandemic. Except, remember, there is a wide margin of error: for predicting deaths in the US. The range is greater than the population of Wilmington, Delaware.

"No model is perfect, but most models are somewhat useful," said John Allen Paulos, a professor of mathematics at Temple University and the author of several books on mathematics and everyday life. "But we cannot confuse the model with reality."

A challenge for modelers is dealing with declining death toll from overloaded public health departments. State data can show wide swings in deaths, but only because one accumulation of reports occurred at a time. The tremendous jumps in deaths in a single day could yield predictions.

Another problem, University of Texas disease modeler Lauren Meyer said, is that most pandemic models, including her own, are based on how influenza works, and that's different from this new coronavirus.

Most models use calculus to account for "things you can't predict," Meyer said. For her, they are simple equations that a person who knows advanced calculus can understand. For the rest of the world, it is Greek. Literally full of sigmas, phis, omegas, and other symbols.

Even with all the uncertainty, "it's much better than shooting from the hip," said Meyer, who is producing iterations of what she calls a COVID-19 "workhorse model,quot; for the Centers for Disease Control and Prevention. of Diseases. "Data-driven models are the best evidence we have."

Due to the large fraud factor, it is smart not to look at a single number, the minimum number of deaths or the maximum, but the confidence range, where there is a 95% probability that reality will fall, mathematician Paulos said. For the University of Washington model, it is 50,000 to 136,000 deaths.

Uncertainty will decrease over time, but it will never actually disappear, just as in hurricane forecasts, when the cone of uncertainty shrinks as the storm approaches landfall, but it remains large.

"Uncertainty is the only certainty there is," Paulos said. "And knowing how to live with insecurity is the only security."


Borenstein reported from Kensington, Maryland.


The Associated Press Department of Health and Science receives support from the Department of Scientific Education at the Howard Hughes Medical Institute. The AP is solely responsible for all content.

Copyright 2020 The Associated Press. All rights reserved. This material may not be published, broadcast, rewritten, or redistributed without permission.


Please enter your comment!
Please enter your name here