How exactly can the case of corona infection be predicted? About pitfalls and pitfalls and why even scientists are wrong.
BERLIN – The outlook for the coronavirus case after Easter was worrying. On March 12, the Robert Koch Institute (RKI) calculated a significant increase in 7-day incidence in Germany. Accordingly, there should be up to 350 new coronavirus infections per 100,000 residents within seven days in mid-April. But this case did not happen.
The B.1.1.7 coronavirus boom was seen as the basis for a massive increase after Easter. Instead, Germany had a nationwide infection rate of 160 on April 15, for example, but why is that? Are such expectations likely to sensitize a population that behaves cautiously as a result? Several experts are now commenting.
|Injury for 7 days in Germany||History|
Corona predictions: how is the infection path prediction model formed?
To understand why it is difficult to accurately predict the future of a case of corona infection, it is important to take a close look at the technology behind the forecast. As epidemiologist Ralph Brinks of Witten-Herdecke University explains, one or more assumptions were made first. Then it is combined into a form. This is then used as a basis for the future course of the epidemic.
The spread of the virus can be well calculated if you assume that people do not change their behaviour. However, as is known, behavior changes for many reasons”, explains scientist Viola Pressman. She developed corona predictions into “alternative scenarios” and heads a research group at the Max Planck Institute for Dynamics and Self-Organization in Göttingen.
Corona: The far-reaching expectations in the future are the same as the weather forecast
Physicist and data scientist Cornelius Romer compares such forecasts of the coronavirus with weather forecasts. But there is a crucial difference. Because when it comes to weather forecasts, people already know that predicting a month in the future may not be completely reliable. It varies with more accurate weather forecasts for the next few days. And the data scientist stresses that “Corona’s predictions should also be presented in this way,” adding: “It helps people understand how reliable they are.”
On Twitter upset the data scientist About what some media reported about Corona models. “Even if you report that the models have an estimate interval, some journalists just sweep that under the table and act as if the model’s predictions should be 100% correct,” Romer wrote, adding in another tweet, adding in another tweet: “We have to wait Even you are certainly not wrong? Then several weeks pass. The costs of an early reaction, even with only 30% certainty, are much less than the benefits. But it seems that politics and the media cannot handle the uncertainty.”
Corona predictions: a publication predicts a turning point in the epidemic
But there are also predictions about Corona that were correct to some extent. In a tweet, data scientist Romer compares one of his predictions about the occurrence of March 21 with the reality that happened. According to Romer, regardless of the impact of Easter, when fewer tests were conducted, his model was correct over a 50 percent time period. To do this, he took into account the following implications: vaccinations, seasonality, b 188.8.131.52 and possible decisions of the Prime Ministers Conference at the Corona Summit.
Viola Pressman and her team also hit the mark when, in a post last summer, they predicted a tipping point at which the coronavirus situation could spiral out of control. The condition was that the numbers continued to rise to the point that health authorities could no longer keep up with registration and tracing. For Brisman, this work “established a fundamental mechanics of reproduction”.
Corona: Scientists criticize the RKI connection
In mid-March, on the other hand, RKI analyzed the net prevalence of variant B.1.1.7: in the Corona forecast, the trend continued into the future, “which we previously observed steadily for eight weeks,” RKI spokeswoman Susan Glasmacher.
In mid-March, the RKI’s focus was on observing the exponential growth of the most contagious British variant B.1.1.7 – and that’s just because: The Brinks epidemiologist emphasized that “the effects of the suppression have not been recorded in any way”. Data scientist Romer adds that the RKI should have communicated more accurately and made clear: “This is not a realistic prediction of what will happen, but a simple model that just shows the effect of the British variable.”
Corona’s long-term predictions in the cycle can affect people’s behavior
Viola Pressman also comments on the diagnosis of the course of corona infection on the part of the RKI. The fact that new coronavirus infections have not increased sharply is due to a change in people’s behaviour. The scientist explained that the other reasons were the advancement of vaccination, testing and possibly seasonality.
For Pressman, the curve was expected to rise more slowly than expected in the simple scenarios. She stressed that it was difficult to predict how slow growth would be. As a result, the prediction of the RKI aura played its role in changing the behavior of people. RKI also takes this line when it indicates a marked decrease in the movement of people during the Easter and holiday holidays as well as closed schools.
Corona Virus: Long-term predictions of the case of corona infection also have problems
At the end of January 2021, the Essen RWI – Leibniz Institute for Economic Research has already dealt with the fact that such long-term forecasts about the course of the Corona pandemic are not without problems. Experts came to the conclusion that predictions can be made only on the basis of existing knowledge. The press release reads: “Therefore, one must always assume that the future will be like the past.” But this is not the case, RWI asserts: “Because the same expectations lead to changes in behavior, so the future is different from the past and the diagnosis is no longer correct.” (Sofia Luther with DBA)