Since AI needs large amounts of data, the challenge with Covid-19 currently is the availability of quality reliable data, says Professor Louis Fourie.
Since AI needs large amounts of data, the challenge with Covid-19 currently is the availability of quality reliable data, says Professor Louis Fourie.

Tech News: Data and Artificial Intelligence in the Covid-19 crisis

By Louis Fourie Time of article published Apr 3, 2020

Share this article:

Artificial Intelligence (AI) has developed dramatically over the past few years and has the potential to assist us with some of the most pressing issues that we experience on Planet Earth. One such a devastating problem where AI can be leveraged to find solutions is the Covid-19 crisis. 

In fact, it was the Toronto-based 2014 start-up Bluedot ( that was one of the first organisations in December 2019 to identify the emergence of a new flu-like outbreak in the Hubei province in China. 

BlueDot uses an AI-driven disease surveillance algorithm that searches more than 100 000 mass media sources in 65 foreign languages, such as news reports, animal and plant disease outbreak reports, official announcements, as well as global airline ticketing and mobile device data. The data sources are scanned for more than 150 different pathogens,  toxins, and syndromes in near-real time.

The results are scrutinised by epidemiologists and then used to give their clients and public health officials, frontline hospitals and airlines advance warning of high-risk areas and possible crises. Using the natural language processing and machine learning capabilities of AI, Bluedot correctly predicted that Covid-19 would jump from Wuhan to Bangkok,  Seoul, Taipei and Tokyo.

BlueDot also correctly predicted the Zika virus outbreak in South Florida and has shown in the last few years that the prediction of disease mobility and outbreaks is possible though the use of AI analytics.


One of the powerful capabilities of AI is machine learning, which entails the ability of AI to learn by itself and to improve itself. It is this capability that enables AI to eventually identify complex patterns in large sets of data, whether text or images. 

If used properly, AI can exceed humans not only in speed, but also in accuracy when identifying patterns in the data that humans might overlook. 

Since AI needs large amounts of data, the challenge with Covid-19 currently is the availability of quality reliable data. 

This problem is exactly what makes our current situation so daunting. Without historical examples to evaluate, it is extremely difficult, if not impossible, to know the best course of action to be taken. But luckily the medical data are currently becoming available as some countries are beginning to recover from the devastating onslaught of the virus.

Several large datasets that can be used by AI to identify patterns are now available. Some of the free datasets are:

The  Covid-19 Open Research Dataset (CORD-19), which contains the text of more than 45 000 scholarly research papers.  Of these articles, 33 000 are full text in particular on the Covid-19 and coronavirus family.

Using natural language processing, AI could generate valuable new insights from the published research. 

The  Covid-19 Research Database containing the latest scientific findings on the coronavirus disease (Covid-19) provided by the World Health Organisation (WHO)

Microsoft Academic resources and their application to Covid-19 research. 

The National Center for Biotechnology Information ( NCBI) PubMed Central (PMC) database of 27 502 scientific articles on Covid-19, Severe Acute Respiratory Syndrome (SARS) and Middle East Respiratory Syndrome (MERS).


A further open-access and free AI tool, Covid-Net, has been made available in the fight against Covid-19. 

Covid -Net uses neural network technology to help scientists to collaboratively develop an AI system that can accurately identify Covid-19 in chest x-rays through the detection of certain prognostic signs. 

Covid-Net is a convolutional neural network – a type of AI that is known for its exceptional capability to recognize images. Two scientists from the University of Waterloo together with the AI-company DarwinAI in Canada developed the system. 

Covid-Net was trained on 5 941 chest x-ray images to identify signs of Covid-19. The x-rays were taken from 2 839 patients with various lung conditions, including bacterial, non-Covid viral infections, as well as Covid-19 infections.

The tool is unfortunately not production-ready, but with the help from scientists all over the world it may soon actively assist in the fight against Covid-19. What makes it unique is that DarwinAI is also working on a function to enable the tool to explain its reasoning, which would make it much easier for health-care workers to use. 


Many people in South Africa, and worldwide, are disturbed by the intrusion into their privacy through the geo-location tracking of their movements and contacts by government. 

And although it may be misused be government in the future when the state of disaster is over, the tapping into digital traces do give health authorities valuable insights into the  behaviour and spreading of Covid-19.

There is thus is simple way people can help to fight Covid-19 beyond just washing their hands – they could donate their data for study by epidemiologists. Many people even wear smart watches and fitness trackers that in principle can detect a fever.  These data could help health care officials to determine where infected people are, as well as whom they have been in contact with, so that preventative actions could be taken. 

Several Asian countries, such as China, Taiwan, Singapore and South Korea required their people to download an app that evaluates them based on their contagion risk and then shares the information with the relevant authorities.

However, privacy conscious countries like Germany are still tracking the spread of Covid-19 through traditional interviews with infected patients to determine where they have been and whom they had contact with. 

When health-care providers and authorities use AI to identify patterns in these huge volumes of cell-phone data, it can be extremely valuable for decision-making and planning. 

Digital epidemiology is an emerging field, which combines big data analysis with methods from medicine, biology, social science, statistics and computer science to model, understand and predict the course of infectious diseases like Covid-19. 

AI could particularly be of value in discovering “clusters” of people, since populations do not fully mix and tend to be partitioned into clusters that rarely interact with other clusters. 

These hidden structures in a community have a significant impact on how infectious diseases spread, as was proven by the spread of Covid-19through the close-knit community of the Divine Restoration Ministries Church on the outskirts of Bloemfontein.

A health care analytics company in the US, Jvion (, uses AI to study 30 million patients in its database to identify people and communities at the highest risk of morbidity and mortality as a result of acute respiratory illness from the Covid-19 virus. 

The creation of the Covid-19 community vulnerability map is based on more than 5 000 variables, which include their medical history, lifestyle and socio-economic factors. 

This enabled Jvion to create lists of people who could be prioritised for hospitalisation or isolated pro-actively. The AI analysis of the data is also used to predict the spread of the virus.

AI could slow down the spread of the coronavirus and extenuate its consequences, but for that epidemiologists need data. However, getting access to personal data will remain challenging in a world where citizens are increasingly distrusting government and where data protection has become a fundamental right.
Perhaps the online platform Open Humans (, asking people to willingly share information as they share it with Google, is the way forward to promote public health and save lives.


AI is not only playing a major role in helping to track the outbreak of the virus, but also to develop a vaccine. Through a deep analysis of the care that every Covid-19 patient receives, AI further assists in determining the best treatment strategies. 

The value of data and AI: There is little doubt that the timely availability of reliable data is extremely important for the effective management of a pandemic such as Covid-19. If big data is combined with AI, it becomes a powerful tool that could assist in limiting the spread of a disease, as well as the mitigation of its consequences.

AI remains a valuable tool to make sense of huge sets of data. When combined with the expert knowledge of humans, it can save human lives. In the current Covid-19 crisis, technology such as AI, machine learning and powerful algorithms, as well as human knowledge, creativity and innovation will all play a critical role in fighting and surviving the Covid-19 crisis.

Professor Louis C H Fourie is a Futurist and Technology Strategist [email protected]


Share this article: