Forecasting AI blog





Artificial intelligence vs coronavirus

Artificial intelligence vs coronavirus

By G. H.


May 6, 2020





Could artificial intelligence help us fight the coronavirus pandemic? has compiled a non-exhaustive list of the scope of AI on Covid-19.

That is debatable! Some say that it has and will have absolutely no use or, at the very least, that it is still lagging behind the pandemic that is affecting the world with the coronavirus. Yet.., not a day goes by without new articles being published on AI and coronavirus. Without taking sides, has therefore decided to gather as much information as possible via publications in the press in an attempt to to see in what way and in what particular field artificial intelligence could bring real added value to human beings in the fight against Covid-19. And its scope is or will be rather broad. In this context, and to prolong the reading of this article, we would strongly advise you to take a look at the publication published on the Council of Europe's Internet site entitled AI and control of the Covid-19 coronavirus and editable in five languages.

Helping caregivers

The contagiousness of the coronavirus being recognized and very strong, Artificial intelligence can first of all relieve the care staff on the front line when dealing with patients. This is concretely the case in the United States and at the Brigham and Women's Hospital in Boston, to cite just one example. There, the robot-dog Spot created by Boston Dynamics has taken over from some doctors to approach patients and carry out an initial consultation without taking the risk of potential contamination. The robot-dog is equipped with an iPad and a radio to allow communication between the nursing staff and the patient. On the official website of the creators of Spot, they specify that it is mainly a tele-consultation. Soon, they hope to be able to carry out a vital inspection remotely with the monitoring of temperature, respiratory rate, pulse and oxygen saturation. In the long term, Spot could also be made available to the cleaning teams for disinfection using a UV-C lamp attached to its back. Hospitals, resuscitation tents and even subway stations could use it to improve the conditions of health care personnel and workers in general.

Predicting and preventing patients with complications

On 30 March this yearhe American and Chinese researchers who have made a name for themselves by announcing that they have developed a tool using artificial intelligence to predict which coronavirus victims will develop severe lung forms. In one of his publications, indicates that the tool, a smart algorithm, analyzed data from 53 patients at two hospitals in Wenzhou, China, to find several indicators that would suggest a patient may be developing a severe form of coronavirus. These indicators include changing in alanine aminotransferase levels, hemoglobin levels and reports of pain. By combining these indicators with other factors, the tool was able to diagnose acute respiratory distress syndrome with an 80% success rate. Advances in research and knowledge of coronavirus will undoubtedly improve this accuracy rate...

Helping deconfinement with tracking

It is fashionable these days to avoid a second wave in all countries that are preparing to deconflict their population en masse. Artificial intelligence to track the coronavirus outbreak. While tracking applications have their supporters and opponents because of their impact on private and public life, on individual and collective freedoms, they can nonetheless provide an accurate mapping of the epidemic in a country, on a continent or around the world. These applications, such as StopCovid or CovidIA in France, do not convince all authorities like the CNIL for example, but they have the capacity to trace the geographical and temporal progression of the epidemic, thanks to a sample of tests carried out on a voluntary basis, and thus to detect and prevent all persons who have been in contact with a patient or a suspect case. They will then receive an alert on their phone to warn them of the risk involved and alert them to go for testing, limit their social exchanges or quarantine themselves in case of proven symptoms. One drawback, however, is that the success of these applications relies on massive adoption by the population (at least 60%), which is not necessarily the case at present in all the states that already use one such as Singapore (20% of the population has downloaded it). However, it should be remembered that the Covid-19 has already contaminated more than 3.5 million people worldwide, 258,000 people have not survived. In France, there are more than 170,000 confirmed cases and more than 25,000 deaths at the time of writing.

Identifying treatment for Covid-19

It was last January and the American monthly magazine... Wired was echoed by him on April 17, 2020. Another scope of application of artificial intelligence on coronavirus, that of helping to identify treatments. The French magazine Science and Nature also addresses the subject in a March 26 publication. CFounder of Flare Capital Partners, which invests notably in healthcare technologies, Michael Greeley explains in particular : "AI allows us to conduct simulations and understand biological evolutions as never before.". And to specify: "Millions of samples are taken and the machine identifies a handful of molecules that might work for a vaccine, for example". In the United States, Wired reports him the story of Peter Richardson, a British pharmacologist, who is said to have found a treatment after teams from the London startup BenevolentAI developed an artificial intelligence that makes it possible to carry out research on steroids by combining data from the pharmaceutical industry and essential information from scientific research documents. And so they discovered Baricitinib, an arthritis drug produced by Eli Lilly Laboratories, which would be effective against coronavirus. Since then, the laboratory in question has launched clinical trials to measure its effectiveness and safety. The results have not yet been released, but this information shows that AI could play a central role in drug research.

Facilitating the discovery of a coronavirus vaccine

Another field of application for AI against coronavirus! Last January, DeepMind launched with great fanfare its 3D protein modeling tool called DeepMind.AlphaFold with a publication in the scientific journal Nature. As part of the search for a vaccine that is not announced, at best, until early 2021, the British artificial intelligence company acquired by Google in 2014 has made its tool available to all researchers under a Creative Commons free license via a download link on the project homepage. The company is already showing results even though some products have not yet been tested in the laboratory. Models of the coronavirus can, however, help to understand how it works and thus save time in the development of this long-awaited vaccine.

Artificial intelligence could be used in the search for a coronavirus vaccine.

Sorting patients before consultation

Allocovid, you know him? Set up since April 27th in France, This artificial intelligence should make it easier to identify all patients with symptoms of Covid-19 and to direct them in the best possible way for their convalescence. Depending on the symptoms, whether they are minor, more pronounced or showing signs of severity, their age, weight or height, this robot will be able to determine whether the patient should simply remain confined, consult their GP or call the emergency room. This vocal bot, which can be reached at 0-806-800-540 in France, will undoubtedly save time for healthcare personnel in the sorting of patients. It is the result of close collaboration between Inserm, the University of Paris, the digital subsidiary of SNCF e-voyageurs and the start-up Allo Media, itself specialised in voice recognition.

Browse research articles on coronavirus

They're not the first people to have tried to unravel the mystery of the coronavirus using artificial intelligence... but Google recently published an article on its blog to announce the launch of its tool COVID-19 Research Explorer in order to help all researchers in their quest for information. Since May 4th of this year, the Mountain View firm has been proposing a semantic search interface to browse more than 50,000 journals and preprints on the subject. The number of articles since the beginning of the pandemic being, since the beginning of the pandemic, consequent with each day of news, the AI can enable researchers to find answers to their questions much more quickly. And thus facilitate, for example, the search for new symptoms or new treatments.

Artificial Intelligence can be used to quickly detect people with coronavirus.

Take the temperature of each employee or traveller

As deconfinement approaches, each company will have to put in place a series of measures to ensure the safety of its employees and visitors. Here again, the example could come from China and Beijing in particular, where some companies are already using facial recognition technologies and infrared sensors to monitor individuals on a large scale. For example, for example.., Qinghe Station in Beijing is already using a system developed by Baidu to automatically photograph each passenger's face and calculate their temperature. This eliminates the need to wait in front of a gantry crane or security guard for an individual temperature reading. In France, some companies have already installed this type of device with thermal cameras at their entrances. In China, all persons who are subjected to the device and whose temperature exceeds 37°3 then go through a second manual check. This reassures the population before going back to work and public transport...