How Artificial Intelligence could help Wales prevent the spread of Covid-19

Keith Darlington, AI consultant and author

As most of the world has now gone into lockdown, the entire scientific research community has gone into overdrive trying to understand the nature of the COVID-19 virus.

Technology, including Artificial Intelligence (AI), is helping with this task. From the beginning, AI applications have been working behind the scenes assisting the limitations of human knowledge in this massive endeavour.  In this article, I briefly describe examples of different ways that this is happening.

Machine learning, as it is known, is the main driving force behind AI. In essence, what machine learning does is to take large amounts of data – called Big Data – and learns to detect patterns in the data. Future outcomes can be predicted, and other useful insights about the data can sometimes be revealed.

But having access to large amounts of data is essential to ensure that a high confidence level can be assigned to these predictions. The machine learning methods can be applied to many problems, as described below.

Tracking

AI has played a role since the very earliest stages of this virus. A Canadian AI company called BlueDot developed an AI program that alerted the world to the virus after the first case was detected in China on December 31^st.

This program was designed to predict infectious diseases and locate and track their spread. It works by combining AI with the knowledge of epidemiologists who identify how and where to look for evidence of emerging diseases.

BlueDot analyses over 100,000 reports daily in many languages and then sends out regular alerts to health care, government, business, and public health clients. The alerts provide brief synopses of anomalous disease outbreaks that its AI program has discovered and the risks they may pose.

But other AI applications have rapidly appeared to monitor not only the spread of information about COVID-19 but also the spread of humans infected and detect humans carrying symptoms. For example, contact tracing smartphone apps were first used in Wuhan to track and trace possible carriers of the virus (see later).

Another AI program is being used in cities in China to detect symptoms of people in bus and train stations as well as other public places where there is a high concentration of people. In this application, AI is combined with sensor detecting temperature measurement technology using computer vision. This technology makes it possible to take body temperature, a key symptom of COVID-19, in a contactless way without affecting normal behaviour.

With this technology in place, those whose body temperatures exceeded the threshold could be located. Doing this manually would be time-consuming and could increase the risk of cross-infection.

Scanning

Testing has become a key issue in the fight against COVID-19. Countries like South Korea and Germany have been seen as successful in handling the virus because of the amount of testing that is done in those countries. Health authorities are keen to increase the numbers being tested but the main testing methods are labour intensive and time-consuming.

But AI is now assisting with other forms of testing, such as x-ray scanning. Various AI programs are now available for chest screening that can highlight lung abnormalities in a chest X-ray scan and provide a COVID-19 risk evaluation much faster than human radiologists.

Resistance

AI has been used in healthcare systems for many years for a range of applications and has encountered some resistance – particularly with regard to use of medical patient data. Having access to medical data raises many sensitive issues of privacy and confidentiality.

This became a contentious matter when the British NHS system failed to comply with data protection rules when it provided 1.6 million patient records to a Google-owned company in 2017 for machine learning analysis.

This, in part, explains the concern with the use of contact-tracing apps, which are already in widespread use in Asia – in countries like China, Hong Kong, Singapore and South Korea. They are also now being used in other parts of the world such as India, Italy, and Israel and development in other nation-states continue apace.

Contact-tracing apps vary in the way they work but generally use the fact that smartphone users whereabouts are detectable and therefore, can monitor close contact with other users. AI algorithms can then determine the risk of cross infection and then alert users of such risks by learning from the data collected.

In the last few days, it has been announced that an app has been proposed by the UK government for use in England, that may also be rolled out in Wales. This app is now being trialled in the Isle of Wight.

Mark Drakeford said on the 4^th May, that there are some difficulties that need to be resolved because of the differences between NHS England and NHS Wales.  As Drakeford said at his daily press conference: “At the moment we are working with the UK government on that app to see if we could make use of it here in Wales”.

It works by using the Bluetooth protocol to identify other smartphone owners who are in close proximity to each other. Thus, if someone develops Covid-19 symptoms, he or she can notify the app of these symptoms. This data can then be uploaded onto the NHS server and people who have been in contact with this person can be notified and possibly given advice – such as requesting self-isolation.

There have been concerns about privacy and the possibility of government surveillance of individuals – i.e., the emergence of the “Big Brother” State – particularly with using this app because of its access to centralised data.  However, the proposed app will only require participants to enter part of their postcode and will not ask them to enter their names. This offers some level of anonymity because the data will be stored under an anonymous ID.

An important distinction between this contact-tracing app and others is that the data will be centralised and stored on the NHS servers unlike some apps used in other countries which work on individual phones rather than collected centrally. The advantages of doing this is that the hotspots are in the country can be detected. And even if there is a low participation rate in the project, of around 20%, then some important insights may be gained into how the virus is spreading.

For all the privacy concerns, many will feel it’s a price worth paying – particularly if assurances are given that this data will be used for this purpose only. Mark Drakeford thought people would be willing to part with some personal freedoms to take part in the project,  and many experts believe that having access to track and trace data is essential.

For example, Dirk Brockmann, an epidemiologist who leads a project tackling the coronavirus pandemic at the Robert Koch Institute in Germany says: “There is a simple way that people can help the fight against coronavirus, beyond washing their hands – donate their data”.

Most people now own smartphones and, if they can be persuaded that it is used for the good of all and used anonymously, then individuals may be persuaded to submit data voluntarily. This is vitally important because according to the University of Oxford’s Big Data Institute, “a contact-tracing app could help stop this pandemic, but 80% of smartphone owners would need to use it.”

Another concern with contact-tracing is that the data may not be a good reflection of the population because the elderly are less likely to use smartphones. But as Mathew Gould, the head of the NHS unit that developed the app, says: “I’m conscious that smartphone use goes down with the more elderly population. This is part of a strategy, so we’re making sure we’re not just relying on the app.”

Conclusions

Fighting this pandemic has been helped by AI adoption from its inception. The use contact-tracing is likely to begin a new phase which we hear a great deal about in the coming months.

Its use is controversial – but may turn out to be crucial in our battle with COVID-19.