AI and data science tool up to battle COVID-19

The COVID-19 pandemic has been unprecedented in its spread throughout the modern world. What can modern technologies, such as data science and artificial intelligence, do to help combat it?

The COVID-19 global pandemic is creating large volumes of data, which data scientists are analyzing to track the disease, guide the response and find treatments. To humans, the data from pandemics can be hard to grasp because there is a long gap between an outbreak happening and visible results in the community – particularly at scale. Data science can be invaluable in crunching these numbers.

Already, many projects are underway using artificial intelligence (AI) and big data analytics to battle the pandemic. They can play a role across the whole lifecycle of the outbreak: from prediction, detection and response, all the way to recovery.

Detection

In late December 2019, Toronto start-up BlueDot spotted an unusual cluster of pneumonia cases around a market in Wuhan, China. This was over a week before the CDC in the U.S. and the World Health Organization issued statements warning of an outbreak of a new influenza-like illness at the same location.

BlueDot, which specializes in infectious disease surveillance, uses an AI algorithm to analyze big data from hundreds of thousands of sources. These include news reports, airline ticketing data, government notices, health data and disease networks to detect the emergence of new infectious illnesses. It uses technology to look directly at data without having to interact with local authorities who may be slower or reluctant to share data.

Spread

In addition to detecting an outbreak, AI can also help predict its progress. For example, BlueDot was able to predict the early spread of the illness from Wuhan to other Asian cities based on airline ticketing data. This has successfully been used before, for example during the Ebola epidemic of 2014, when researchers in the U.S. created a travel-census model to predict exactly where in Texas a case would be found.

Mobile phone data can play a key role in tracking the movement of people to help identify where the disease is likely to spread. For example, location statistics can help analyze the spread of the disease and help in allocating resources according to population distribution.

AI-driven predictions of disease spread can then guide public health authorities in their decisions for resourcing and actions required in certain locations. For example, big data analytics can cross reference disease data against high-risk senior residents down to postcode level and the incidence of factors such as diabetes or obesity. This allows them to identify where extra intensive care beds are likely to be required, for example.

Managing the pandemic

Social distancing is being used worldwide as the key weapon to minimize onward transmission of the disease. One of the difficulties encountered with COVID-19 is knowing when and how the confinement period can be interrupted. Lifting the confinement too early could have the consequence of triggering a second wave even more uncontrollable than the first.

Didier Gaultier, Data Science & AI Director at Business & Decision, Orange Group, says that we are currently missing the large-scale data on COVID-19 required for data science and AI to give accurate results. For example, researchers suspect that blood type plays a role in the risk factors for developing severe forms of COVID-19, but studies and data are still missing, and this is probably not the only factor involved.

WHO considers testing to be vital in combating the spread of the disease, so there is research underway using AI to identify COVID-19 patients from other sources, such as CT scans. Machine learning is already used to identify all types of illnesses, from cancer to eye diseases, so this is a route with some potential, even if CT scans are unlikely to be useful as an early warning signal of the illness.

Data science can play a central role in analyzing the large-scale testing of people by linking these results with anonymized health characteristics of hospitalized patients. This will allow us to understand the key risk factors and better protect people with the greatest risk. Because it is a new disease, these are not yet fully understood. The more data there is, the more accurate these predictions can be, and the better the pandemic can be managed. This will also allow us to make better judgements of when and how social distancing should be lifted.

AI is also being used to accelerate drug development to treat COVID-19. For example, Google’s Deep Mind AI system is being used to identify characteristics of the virus that may help to understand how it functions. This would be useful information in working out what treatments to pursue. Others include UK-based BenevolentAI, which is using AI to identify promising existing treatments for other illnesses that could be effective in treating COVID-19.

Recovery

Once the outbreak has been contained, public health authorities can use AI and data science to make decisions about how to deal with outbreaks in the future. The data that is collected from this pandemic will be invaluable in understanding how best to deal with future outbreaks. It will allow authorities to test different scenarios and outcomes to make data-driven decisions on the best actions to take in the future.

Artificial intelligence can also help identify where the next outbreak will come from and what form it will take. While scientists had expected the next global pandemic to be influenza-based, the fact that COVID-19 is from a coronavirus is not a surprise. In fact, it is closely related to the SARS and MERS epidemics.

What unites influenza, coronaviruses and other infectious diseases, such as Zika and Ebola, is that they have all made the jump to humans from animals. In fact, the CDC in the U.S. believes that three-quarters of new diseases in humans have originated in animals. Many believe that encroachment by humans onto previous virgin land will create a risk for more of these diseases.

Early knowledge about a disease outbreak can play a vital role in improving the response to the illness. This makes global disease surveillance an important part in the battle against future pandemics. AI is playing an increasingly important role in this activity. It can help researchers analyze global data about known viruses, human activity, disease modeling, visualization and mapping to predict where the next pandemic will arrive and the impact it will have.

Infectious diseases have always co-existed with humankind, and while modern life has been instrumental in accelerating this pandemic, we have never been as well equipped to deal with it. Understanding this pandemic is vital in combating it, and of course, we still have much to find out about COVID-19. The more data we collect, the better data science and AI will be able to help us.

Find out how Orange is mobilized to ensure continuity of service for all customers, in France and around the world.

Pierre-Louis Biaggi
Pierre-Louis Biaggi

I am a member of the Management Committee of Orange Business Services and head the Digital and Data entity. With the expertise of 3,500 consultants, developers and data scientists, as well as an open ecosystem of partners, we enable businesses to harness the power of data and digital technology to innovate and reinvent their activities.