How can artificial intelligence help overcome the crisis generated by the COVID-19?

ARTIFICIAL INTELLIGENCE HAD DETECTED THE EPIDEMIC SEVERAL DAYS BEFORE THE WHO.

Last January, we learned that an AI from BlueDot, a Canadian startup, had in fact already anticipated the Covid-19 epidemic. At the end of December, the algorithm would have detected the appearance of the coronavirus in the province of Wuhan and correctly predicted the evolution of the virus towards other cities in Asia. How did this system get ahead of the World Health Organization? The BlueDot algorithm uses natural language processing and machine learning techniques and continuously scans for newsletters in many languages, official statements, reports of new plant and animal diseases as well as publications on forums and blogs that relate unusual events. By collecting this data, the algorithm can then detect the signals of an epidemic and their location. Subsequently, it is even able to determine where and when individuals infected with the disease go using data from ticket offices of global airlines. Epidemiologists will then check the relevance of the analyzes of artificial intelligence and will validate or not the conclusions obtained. Unfortunately, the BlueDot alerts were not taken into account by the health authorities who had received the startup's conclusions.

In the United States, the Harvard University School of Medicine publishes a world map that tracks viruses in real time: HealthMap . Unsurprisingly, the coronavirus has been in first place in recent weeks. As with BlueDot's technology, NLP bots scour the web for signals about viruses: social networks, official press releases, news sites, etc.

AI, A TOOL TO FIGHT CORONAVIRUS IN MEDICINE.

In early March, Alibaba, the second largest high-tech company in China, developed an artificial intelligence algorithm to diagnose Covid-19 with a success rate of 96%. This AI model has been trained using thousands of lung scanners from Covid-19 patients. 15 seconds is the time it takes to diagnose a patient's coronavirus while it takes 5 to 15 minutes for a radiologist. Significant time savings at a time when hospitals are overcrowded. This system has been deployed in around 100 establishments in China to help doctors make their decisions.

This Wednesday, March 25 in Russia, researchers from the Gero company would have identified several drugs that would prevent the multiplication of Covid-19 in human cells. Thanks to the development of computing power and the emergence of deep learning, thousands of drugs and their effects on the virus were analyzed very quickly. Six drugs were then said to be effective in suppressing the proteins responsible for the reproduction of the virus. Laboratory tests are still in progress. In addition, this Monday, March 30, Russia made available a new screening test that can detect the virus in 90 minutes. 100,000 people will then be tested in seven days.

Elsewhere in the world, artificial intelligence is also used to help citizens. In Spain, Raquel Yotti, director of the Madrid Institute of Health, announced at a conference the investment in four robots to automate COVID-19 screening and thus pass 80,000 tests per day, four times more than before. On the United States side, the publisher IPSoft has created a conversational agent named Amelia to converse with American citizens and assess their risk of infection based on the symptoms and potential risk factors. After the conversation, Amelia will advise calling a healthcare professional if there is a risk of infection, or will notify that the risk is low.

DeepMind, Google’s artificial intelligence division, has also attempted to thwart the Covid-19 epidemic. To do so, she used her AlphaFold tool to generate 3D models of proteins from the disease virus. They were distributed free of charge to the scientific community. The algorithm was trained using proteins from the Protein data bank for five days. The company says this was possible thanks to "free modeling", which allows AlphaFold to predict plausible protein structures without needing to rely on known structures. A real asset considering the urgency of the situation given that the current coronavirus presents some novelties.

THE DEPLOYMENT OF ARTIFICIAL INTELLIGENCE. TO THE DETRIMENT OF INDIVIDUAL FREEDOMS.

In China, individual freedoms and artificial intelligence do not always mix. With the arrival of Covid-19, the country has widely deployed its AI algorithms to detect people potentially infected with the virus. For example, thermal cameras have been developed in public spaces to detect if a person has a fever, the main symptom of the virus. 200 people per minute could be identified in this way. The facial recognition systems will then recognize who the feverish people are, whether they wear a mask on their face or not; these systems, more and more efficient, are now able to recognize a person even with a mask on the face thanks to the calculation of the space between the eyebrows or the size of the forehead for example.

Alibaba went even further by using its Alipay payment application - already widely used by Chinese people to pay electronically - to test the risk factors of its users. How does it work ? A declarative questionnaire informs your state of health, that of your loved ones and asks you very specific questions about your recent trips. Depending on the responses, the application displays a green, yellow or red color code. If it is green, you are entitled to travel, if it is yellow, you will be asked to stay at home for 7 days and if it is red, you must remain in quarantine for 14 days. In order to access certain cities or certain public buildings, this color code served as a control point. At the same time, drones identified people who were moving without wearing a mask. The same type of technology monitored publications on social networks in order to track people complaining about their condition and cross-check this information with that relating to public transport. The security vs freedom debate has only just started here, it does not exist in China.