The global coronavirus pandemic has led to unprecedented scenarios in almost all aspects of life, throughout the world. A battle to reduce the spread of infection, keep mortality low and maintain both economic activity and public order requires critical decision making in highly ambiguous spaces.

This pandemic is also the first to create an enormous amount of data related to those many facets of life. As machine learning (ML/AI) is essentially the practice of teaching systems to learn with examples in data, it is a potent tool in assisting with much of the necessary critical decision making ahead of us.

This person does not need AI to show them how to wash their hands. Photo by Jasmin Sessler.

There are many ways AI can assist in dealing with both the direct and indirect consequences of the coronavirus pandemic. In order to gradually reduce social distancing measures, it is essential to identify and track cases of the virus as quickly as possible. Many countries have a backlog for testing; this causes a lag in reported cases. AI can help through nowcasting the number of cases to give a more accurate real time estimate and help hospitals better manage their resources. South Korea, for example, has used AI in this and a number of other ways to great effect.

Epidemiological models powered by AI can predict the spread of the disease in social systems to identify and prevent clusters of infection, particularly in vulnerable communities. This can in turn inform policy making, providing agility in dealing with spikes and the implementation of more granular social restrictions. The power of AI to cut through noise in data is beneficial here; with so many signals it can be hard for a human to find the ones that really matter. Given enough data, AI can do this effectively.

AI can also help in the critical task of finding treatments. Computational biology can help us understand the virus by providing information related to its structure and function. DeepMind recently published information related to the potential structure of several key proteins of which the virus is composed. By understanding this, it increases the possibility of finding targets for pharmaceuticals that could disrupt how the virus replicates in the body; a treatment. A recent publication in Cell demonstrated the power of AI in detecting novel treatments for bacterial infections; a field in need of innovation. While bacterial and viral infections are very different, the same approach to drug discovery may work in each case.

Making decisions is hard, particularly when you are lost in a forest. Photo by Jens Lelie.

Bad actors do not waste a crisis, and online misinformation has been rampant from the beginning of this pandemic. AI can help discriminate fake news and other misinformation from authoritative sources, ensuring people have the information they need to stay safe and well. This is particularly important when public health issues become politicized and political opinion starts to overshadow medical fact.

When a vaccine is discovered, AI could assist in the monumental task of ensuring the billions of individuals who will need it across the world receive it as quickly and safely as possible. The use of AI in supply chain management is well established; using data related to the production, transport, storage and delivery of goods and services to optimise those processes. Currently, there is a global shortage of ventilators, masks and other protective equipment. AI can also aid in ensuring the supply chain for these products is as efficient as possible.

While no panacea, AI has a role to play in the battle against COVID-19. With many organisations working together to fight this global pandemic, I hope that the lessons and innovation that emerge from this fight will help us to deal with future public health crises in years to come.

– Colm