New research by the VUB (Vrije Universiteit Brussels) proofs that Artificial Intelligence (AI) can successfully support policymakers against epidemics such as COVID-19. AI can potentially combine epidemiologic models with complex decision-making models that are capable to take into account factors such as human behavior. However, predicting this human behavior is incredibly difficult (Subrahmanian and Kumar, 2017). Therefore, the question arises: How can AI combine epidemiologic models with complex decision-making models to assist policymakers?
Reinforcement learning
Reinforcement learning is an AI technique that became famous after the world champion of the GO-game was defeated by the AlphaGo-program from Google Deepmind in 2017. Pieter Libin examines the epidemiological applications of this AI technique in his research (press.vub.ac.be, 2020). Reinforcement learning is particularly suitable to track complicated interactions. It includes cultural, behavioral, and societal factors to successfully determine prevention strategies to reach specified objectives in an epidemic.
The battle against pandemics with AI efficiency
AI allows proposing a targeted approach to actively combat an epidemic such as COVID-19 (press.vub.ac.be, 2020). The models can learn an optimal strategy that can for example consist of the mandatory wearing of mouth masks, and the vaccination of specific target audiences (given that there is a vaccination) (www.bruzz.be, 2020). Besides reinforcement learning, there are other machine learning (ML) techniques that can be useful against the battle of epidemics. ML techniques in combination with epidemiological and statistical models can predict how fast an epidemic can spread. Therefore, intensive care units can be determined that are extremely vital in hospitals. This way medical institutions are more prepared to make the necessary adaptions in their infrastructure in the eye of an epidemiologic storm. Additionally, Artificial Intelligence can assist in developing new medicines. Specifically, pattern recognizing techniques can make active connections out of bio-Informatica databases to contribute to the design of a new medicine against the virus.
Current AI initiatives against COVID-19
The VUB has already come up with AI initiatives on both the university and international level. Together with the medical experts of the Universitair Ziekenhuis Brussel (UZ Brussel), the VUB set up two different AI-driven projects. On the one hand, an analysis was made about how AI can contribute to more accurate long scans. On the other hand, there is the development of an AI application that doctors in intensive care can use to assist them in their decision making (press.vub.ac.be, 2020).
To conclude, AI techniques such as reinforcement learning and ML techniques in combination with epidemiological and statistical models can propose a targeted approach to combat epidemics. This gives policymakers grounded evidence to construct better policies against epidemics.
This topic is clearly of high interest to me. However, I was wondering what do you think about the ethical implications of the use of AI in the battle against COVID-19? Do you know other AI initiatives against pandemics such as COVID-19?
—
[References]
press.vub.ac.be. (2020). VUB zet AI in tegen COVID-19. [online] Available at: https://press.vub.ac.be/vub-zet-ai-in-tegen-covid-19 [Accessed 7 Oct. 2020].
Subrahmanian, V.S. and Kumar, S. (2017). Predicting human behavior: The next frontiers. Science, 355(6324), pp.489–489.
www.bruzz.be. (2020). VUB onderzoekt het gebruik van AI om epidemieën in te dijken. [online] Available at: https://www.bruzz.be/wetenschap/vub-onderzoekt-het-gebruik-van-ai-om-epidemieen-te-dijken-2020-05-29 [Accessed 7 Oct. 2020].
Hi Thibault,
Very interesting article considering that we are in the middle of this very special pandemic.
I can see you are relating to some of the major benefits of Artificial Intelligence such as ‘better decisions’ as it helps to define an optimal solution (as you mentioned) consisting of wearing a mouth mask and the vaccination of specific target audience. However, do you think it could have helped to be more accurate on identifying the virus and creating a vaccine itself? If yes, how come that we are still lacking a vaccine available for all? Furthermore, do you think it could help to more accurately determine who needs to be tested and/or quarantined?
Additionally, to answer one of your question which was: what do you think about the ethical implications of the use of AI in the battle against COVID-19?
Indeed, they are some ethical challenges with the use of AI in the battle against Covid-19 such as bias, bad algorithmic decisions threatening human rights and safety, misinformation and privacy concerns. However, I believe that if AI is used with a certain amount of accountability, responsibility and transparency, it can very much act as a factor of positive change. Indeed, it would help, as you mentioned, to determine ‘how fast an epidemic can spread’ and help ‘medical institutions to be more prepared and to make the necessary adaptions in their infrastructure’. Which would in turn make the society a better and safer place.
Cheers,
Thanks again for the interesting article!
Dear Sacha,
Thank you for the wonderful input. Highly interesting points you raised. To give an answer to the “However, do you think it could have helped to be more accurate on identifying the virus and creating a vaccine itself? If yes, how come that we are still lacking a vaccine available for all?”. I do not think that AI has the capabilities to create a vaccine by itself as a vaccine is a complicated matter that has to go through different layers of experiments and testing (which can be regarded as data points). However, once vaccines (data points) will become available, I do believe that AI can help to decide which vaccines are the most effective by aggregating these large sets of data points to make more accurate decisions.
To elaborate on: “Furthermore, do you think it could help to more accurately determine who needs to be tested and/or quarantined?”, I do think AI has the capability of this. As highlighted by a fellow scholar (Emma van der Molen) in another blogpost the corona track application in the Netherlands became available to track people who tested positive. Hence, in combination with the accurate implementation of these types of applications, AI will be able to track people that need to be tested and/or quarantined. Nevertheless, ethical issues (more specifically privacy) play once again a role here.
For a read on Emma’s article here is the link: https://digitalstrategy.rsm.nl//2020/10/07/23591/
Cheers,
Thibault