The Coronavirus Pandemic has become life changing for all of us wherever we live in the world, the disruption to businesses and our lives unprecedented and profound. We are all facing the challenges of self-isolating, homeworking, home schooling, food shortages, illness and for some loss of our friends, colleagues or loved ones.
But never has humanity been so together in fighting a cause, with technology playing a massive part in our continued connectedness with family, friends, and work colleagues while being asked to keep so far apart.
In the press we hear daily about the role Artificial Intelligence is playing in helping and potentially solving the current Coronavirus outbreak, but how much of this is real and what impact will this have on our privacy and civil liberties?
Whilst this article is not an introduction to Artificial Intelligence it is important to understand the distinction between Artificial Intelligence (an overall term for many different areas) Machine Learning (gaining insights from data without explicit programming) and Deep Learning (a subset of Machine Learning that uses multi layered artificial neural networks)
Sounds complicated? Not really, if you have asked questions of Alexa or Siri then behind the scenes, they are using Deep Learning techniques to understand and respond to your questions.
There are many areas where Artificial Intelligence and the Coronavirus intersect, such as healthcare, the economy, the future of work, and the Internet.
For the purposes of this article I have focused on healthcare where there are the most diverse use cases. Even just within healthcare there a number of areas where artificial intelligence is having an impact on solving the Coronavirus pandemic, such as:
For some years social media has been used to track and predict the annual flu epidemics, so to some extent forecasting virus outbreaks is nothing new. However, track and trace is different and new. The term itself sounds slightly ‘big brother’ and in many countries has led to questions around civil liberties.
Countries are employing different approaches to track and trace, but the basic principle is to determine your current and past locations and those people with whom you have been in contact. It is thereby possible to assess your exposure risk and likelihood of spreading the virus.
In China, they have used a combination of CCTV image recognition in conjunction with a mobile phone app and each citizen has a red, amber or green QR code on their mobile phone which determines whether they should be in quarantine or be allowed to travel outside of their province.
This uses a combination of Artificial Intelligence techniques, one being image recognition including the use of Deep Learning techniques to recognise the person. China recently claims it can accurately identify a person even if they are wearing a face mask! It uses Machine Learning techniques to analyse the information (‘big data’) from the millions of citizens and their movements.
The UK Government is in the process of implementing a similar app called ‘NHS CV19’ for contact tracing, the intended app is said to be an ‘opt in’ system for smartphones only with the government working with privacy experts to ensure data is anonymised, safe and only used for the duration of the virus outbreak. However, already concerns are being raised as you can see in this letter from ‘Women Leading in AI’ and also Google and Apple are recommending more stringent privacy measures.
One of the terms that has become part of our ‘new normal’ vocabulary is ‘social distancing’ or keeping a safe space between yourself and other people outside of your home in order to prevent possible spread of the virus (we are told to keep at least 2 metres apart from each other).
A company called Landing Artificial Intelligence, can offer analysis of CCTV video footage in real time to highlight those people not keeping two metres apart from each other, demonstrated in the picture demonstrated in the picture. Whilst this analysis undoubtedly could help with ensuring we keep social distance from each other, it will also be seen by many as an invasion of privacy. This analysis uses Deep Learning techniques and specially convolutional neural networks.
Artificial Intelligence is already helping with diagnosis of Coronavirus in several ways, from the obvious to the less obvious. Early in March, DeepMind (part of Google) released a computational model of predictions of the protein structures associated with Coronavirus to help medical professionals develop tests for the virus.
Mentioned previously in this article were convolutional neural networks which are used for image recognition. Several solutions using this type of neural network have been developed to help with diagnosis. One such solution analyses chest scan images, relieving some of the workload from radiologists who had to review and prioritise every scan. This is with the hope that eventually there may be the ability to predict the need for a patient to be ventilated.
Another diagnosis solution is the example of how existing thermal imaging technology has been altered to help detect people with a fever. However, this is also an example of some of the misrepresentation around coronavirus solutions. Whilst it is not for me to comment on the accuracy of claims or counter claims, this is an example of the potential to use the coronavirus pandemic for profitable gains. Be reassured that the theory and application of Deep Learning algorithms for detection in this area is sound.
The final example of diagnosis assistance is using recurrent neural networks (RNNs) for analysis of cough sounds. RNNs are like CNNs, but instead of image analysis they are used for sound analysis; and mostly used with smart devices like Alexa, Siri and Google Home to recognise our voices and commands.
Healthcare professionals have said coronavirus patients have a distinctive cough that sounds different to other illnesses. Using recordings of coughs from patients who do, and do not have Coronavirus, it has been possible to train a Deep Learning algorithm to identify from a cough who has the virus. Whilst this is still in early stages of development this could significantly help with the diagnosis of the virus.
One of the best ways to find possible treatments or cures for coronavirus is using Artificial Intelligence. Artificial Intelligence is particularly good at sorting through large quantities of data to find solutions that might work. Many of the big tech companies are making available their Artificial Intelligence and Cloud resources to scientists who are looking for solutions. However, there are challenges still with some of the large pharmaceutical companies who are less willing to share their valuable data which could contribute to finding a cure. Find out more here.
The coronavirus crisis is global, it does not discriminate age, gender, race or religion. Therefore, when we are looking for treatments or cures, we also need to ensure that the solutions work for everyone. Often the data sets being used come from limited or underrepresented populations. An example of that being the Optum algorithm referenced in this article.
Possibly the most exciting, but furthest off opportunities are in drug design using ‘In Silico’ clinical trials. Rather than using animals or humans for testing, the testing takes in computer simulation at high speed, on virtual patients. The computing power and algorithms required for such computations are at the limits of current classical computers, but these limits will not be in the era of Quantum Computers. Quantum Computing will allow for sequencing and analysing DNA in real-time. For more information on ‘Coronavirus and Quantum Computing, why it matters now more than ever’ see here.
Like many of the Artificial Intelligence technologies mentioned, chatbots have been around for some time; whether you use Alexa, Siri or when you are browsing the Internet and are asked if you would like to chat with someone about your purchase or a service. Chatbots are just computer programs that try to simulate our conversations and can quickly give programmed answers to questions.
It is probably no surprise that some of the first applications of Artificial Intelligence for Coronavirus are based on simple Frequently Asked Questions (FAQs), and these are used to provide information to help people make decisions on whether they need medical attention, and prevent health systems and professionals from being overwhelmed. For instance, Microsoft has worked alongside the US Center for Disease Control and Google has launched their chatbot service providing information and guidance regarding symptoms, making it Open Source so that others could easily build similar services tailored to their needs and audiences.
Chatbots use a number of Artificial Intelligence technologies, but are primarily using Natural Language Programming (NLP), a Deep Learning technique that interprets and understands our ‘natural language’ and responds similarly with natural sounding sentences. Many chatbots can use language translation services, so although the chatbot may be written in English, it can be translated into and understand other languages without the need for multiple versions of the chatbot. Chatbots can learn from our interaction, so can improve accuracy and responses over time.
I have avoided any claims that Artificial Intelligence alone will solve the current pandemic, it will not. The use of Artificial Intelligence to work on Coronavirus problems and solutions is already widespread. The usefulness of all applications of Artificial Intelligence comes from its combination with human intelligence and expertise.
As the world rapidly tries to solve this current crisis, we need to be aware of the potential impact on our privacy and human rights. Even in these exceptional times these should not be compromised. The World Economic Forum recently published an article on how governments must build trust in Artificial Intelligence to fight the Coronavirus. In these uncertain times, we should all still try to keep well informed on the choices and decisions our governments are making on our behalf and be willing to raise our concerns where necessary.
Nigel is a Global Speaker, Influencer and Advisor on Artificial Intelligence, Innovation & Technology (Ranked amongst the top AI Influencers in the World) — ex European Chief Technology Officer at Microsoft, but now an independent voice on Artificial Intelligence and Machine Learning, and Co-Founder of We and AI.