How can Big Data analysis help defeat the coronavirus and how can machine learning technologies allow us to analyze a huge amount of data? Answers to these questions are being sought by Nikolai Dubinin, host of the Industry 4.0 Youtube channel.
Big data analysis is one of the most powerful ways to track the spread of the virus and defeat the pandemic. 160 years ago, a story happened that clearly showed how important it is to collect data and quickly analyze it.
Map of the spread of coronavirus in Moscow and the Moscow region.
How did it all begin? 1854 London’s Soho area is hit by a cholera outbreak. 500 people die in ten days. No one understands the source of the spread of the disease. At that time, it was believed that the disease was transmitted due to the inhalation of unhealthy air. Everything changed doctor John Snow, who became one of the founders of modern epidemiology. He begins to interview local residents and puts all identified cases of the disease on the map. Statistics showed that most of the dead were near the Broad Street standpipe. Not air, but water poisoned by sewage caused the epidemic.
Tectonix’s service shows, using the example of a beach in Miami, how crowds can affect the spread of epidemics. The map contains millions of pieces of anonymous data with geolocation coming from smartphones and tablets.
Now imagine how fast the coronavirus is spreading across our country after a traffic jam in the Moscow metro on April 15. Then the police checked the digital pass of every person who went down to the subway.
Why do we need digital passes if the system cannot cope with their verification? There are also surveillance cameras.
According to Grigory Bakunov, director of technology dissemination at Yandex, the face recognition system that operates today recognizes 20-30 fps on a single computer. It costs about $10. At the same time, there are 200 cameras in Moscow. To make it all work in real mode, you need to install about 20 thousand computers. The city doesn’t have that kind of money.
At the same time, on March 15, offline parliamentary elections were held in South Korea. Turnout over the past sixteen years was a record – 66%. Why are they not afraid of crowded places?
South Korea has managed to reverse the development of the epidemic within the country. They already had a similar experience: in 2015 and 2018, when there were outbreaks of the MERS virus in the country. In 2018, they took into account their mistakes of three years ago. This time, the authorities acted especially decisively and connected big data.
Patient movements were monitored using:
recordings from surveillance cameras
credit card transactions
GPS data from citizens’ cars
Mobile phones
Those who were in quarantine had to install a special application that alerted the authorities to violators. It was possible to see all the movements with an accuracy of up to a minute, and also to find out if people were wearing masks.
The fine for violation was up to $ 2,5 thousand. The same application notifies the user if there are infected people or a crowd of people nearby. All this is in parallel with mass testing. Up to 20 tests were done in the country every day. 633 centers dedicated only to coronavirus testing have been set up. There were also 50 stations in parking lots where you could take the test without leaving your car.
But, as science journalist and creator of the N + 1 science portal Andrey Konyaev correctly notes, The pandemic will pass, but personal data will remain. The state and corporations will be able to track user behavior.
By the way, according to the latest data, the coronavirus turned out to be more contagious than we thought. This is an official study by Chinese scientists. It became known that COVID-19 can be transmitted from one person to five or six people, and not two or three, as previously thought.
The flu infection rate is 1.3. This means that one sick person infects one or two people. The initial coefficient of infection with coronavirus is 5.7. Mortality from influenza is 0.1%, from coronavirus – 1-3%.
The data is presented as of the beginning of April. Many cases go undiagnosed because the person is not tested for coronavirus or the disease is asymptomatic. Therefore, at the moment it is impossible to draw conclusions about the numbers.
Machine learning technologies are the best at analyzing a huge amount of data and help not only track movements, contacts, but also:
diagnose coronavirus
look for medicine
look for a vaccine
Many companies announce ready-made solutions based on artificial intelligence, which will automatically detect coronavirus not by analysis, but, for example, by X-ray or CT scan of the lungs. Thus, the doctor begins to work immediately with the most serious cases.
But not every artificial intelligence has sufficient intelligence. At the end of March, the media spread the news that a new algorithm with an accuracy of up to 97% could determine the coronavirus by X-ray of the lungs. However, it turned out that the neural network was trained on only 50 photographs. That’s about 79 fewer photos than you need to start recognizing the disease.
DeepMind, a division of Google’s parent company Alphabet, wants to completely recreate the protein structure of a virus using AI. In early March, DeepMind said its scientists had come to an understanding of the structure of proteins associated with COVID-19. This will help to understand how the virus functions and speed up the search for a cure.
What else to read on the topic:
- How Technology Predicts Pandemics
- Another coronavirus map in Moscow
- How do neural networks track us?
- The post-coronavirus world: Will we face an epidemic of anxiety and depression?
Subscribe and follow us on Yandex.Zen — technology, innovation, economics, education and sharing in one channel.