AI against Covid-19 Challenge
How can we predict the growth of new infections? How do we know if our measures against COVID-19 are working? What are the implications for how the virus is spreading?
How can data help fight a global pandemic?
We were all watching the news closely. By March 2020 some of us started to make calculations, visualizations to try and get a grip on what’s happening. What may have started as a way to gain a feeling of control over the situation, quickly developed into a serious effort.
At FruitPunch AI we want to harness the data-science know-how of the FruitPunch AI community & network and try to predict some key-metrics going forward. At what point will our healthcare system break? What is the impact on global health of this economic downturn? We’ve decided to develop a system that uses artificial intelligence to analyze how effective the Dutch government’s policy is in combating Coronavirus infections, how to better predict the spread of the infections and the impact on the economy.
Our challenge partners
GOAL: A crystal ball for Covid-19
Over the next two months, the platform from the AI engineers will collect data that will provide insight into how citizens have behaved in response to measures taken to stop the disease. In addition, the platform is to analyze the impact of the pandemic on the economy. It will do this by comparing the increase in the number of Corona patients with the fall in stock market prices, among other things. On this basis, it is possible to assess the probability of a recession as a result of the pandemic. And predict how deep it might become.
The greater economic effect and its snowball effect on wellbeing
When we started we saw two big issues that required more attention:
- The burden on the healthcare system & its breaking point
- The greater economic effect and its snowball effect on wellbeing
Both of these challenges require one first input, a more accurate prediction of the infection spread. It requires either an exact simulation of this world down to the up-quark or a deep insight into how a coronavirus (or a flu virus) spreads in specific areas. With historical, fine-grain data of airborne diseases spreading, we can predict the further spread of covid-19.
We have partnered up with Omdena to take this challenge to the next level. We’ll be analyzing the effectiveness of different policies governments have instated. More information on this project can be found here.
Did you know
- infection rate (R0) of the common flu is 1.3, giving 146 infections after 20 contaminations, covid-19 has an R0 of ~2.5, resulting in *36 million* infections in the same timespan!
- the death rate of the flu is 0.1% and of covid-19 1.6%, resulting in current estimates of 10 to 100 times more deaths than the flu globally, aka 5 to 50 million people!
- biggest threat now: a collapse of the healthcare system due to overloading – there’s only beds for about 0.28% of our population, we need to slow the infection now.
- solution: a swift and common community response is the only way we can contain this virus, by canceling events, limiting social contact (especially with at-risk groups like your grandparents or other people’s children) and of course abiding by the hygiene guidelines.
An AI Crystal Ball for COVID-19?
by D'vorah Graeser, Founder & CEO at KISSPlatform Europe BV
How do we know if our measures against COVID-19 are working?
by Robert van Dijk, Intern at Broad Institute of MIT and Harvard
Data scientists take on corona data to predict growth of new infections
By Edwin van den Heuvel, TU Eindhoven professor & statistician