While the COVID-19 crisis spread around the world, meticulous models forecasted the short-term future state of the outbreak and provided critical intelligence for decision-makers on ways to mitigate and contain the disease.
But these data points shifted as public interventions — like social distancing — begin to take effect, and a new study published in Science shows how changes to the coronavirus' growth rate matched well with publicly announced intervention measures, in Germany.
Modeling COVID-19 outbreak interventions in Germany
In the early days of the COVID-19 outbreak, ongoing and reliable short-term forecasts were crucial material for health officials to estimate what medical requirements and logistical efforts were needed. This also helped authorities advise and inform the general public amid the expanding crisis, according to the Science study.
The initial phase stipulated three main tasks to maximize intelligence for crisis mitigation. The first goal was to establish central epidemiological parameters like the reproduction number — which represents how many more people a virus in one infected person is likely to infect via transmission. This was used for short-term forecasting.
The second was a mass effort to simulate the effects of varying interventional strategies — like comparative battle plans — to minimize an outbreak. The third entails estimation of the real-world effects of measures taken to make rapid adjustments and adapt short-term projections.
In other words, the three pillars of COVID-19 outbreak mitigation are interwoven, like a medical emergency feedback-loop.
Magnitude and timing of COVID-19 interventions matter
It's important to note that the magnitude of a community's outbreak is proportional to the lateness of intervention implementation.
However, accomplishing all of these is hard because of large margins of statistical and systematic errors that happen during the early stages of an epidemic — when case numbers are low, and difficult to project into the future with simulations.
As with any real-world phenomenon, the more data available, the more accurate projections into the future become. But in the case of the coronavirus crisis, this means intervening steps may not take place until a delay — due to margins of real-world error.
Germany's social distancing effectivity modeling improved
To get around these complications, all available prior knowledge must be integrated into the collective modeling efforts to minimize uncertainty factors. These involve basic mechanisms of how the disease is transmitted and knowledge about how and when recovery happens for infected patients.
The models depicting the effectivity of mitigating interventions like social distancing efforts on COVID-19 outbreaks become better with time as daily reports of new cases either match projected outbreak models, or don't.
And when they don't the models need to be changed — these are called "change points" — to improve model accuracy, and represent moments when the projected effects on the outbreak are "updated" to reflect newer and newer data on how effective measures like canceling events, closing schools, and other steps are as time goes on. And according to the study, models with two or three change points displayed optimal predictive efficacy.
While there's no such thing as a perfect response to a global pandemic like the COVID-19 outbreak, it's perhaps comforting to know that some countries have applied their full scientific capacities — and learned how to manage the coronavirus crisis from verified intelligence.
We have created an interactive page to demonstrate engineers’ noble efforts against COVID-19 across the world. If you are working on a new technology or producing any equipment in the fight against COVID-19, please send your project to us to be featured.