Can AI/ML prepare us better to deal with global crises in post-COVID world?

The world has been caught grossly unprepared to meet the extraordinary challenges posed by the Covid-19 outbreak. The extreme ramifications of the global crisis have already begun to unfold. Businesses across the world are severely impacted and the global economy is heading towards what’s said to be the worst global recession since World War II.

Over the last few years, enterprises have been leveraging data science, Artificial Intelligence (AI) and Machine Learning (ML) to draw meaningful insights and make informed decisions. In the light of the current pandemic, we are forced to wonder if things could have been any different. Could any travel/hospitality data model have predicted such severe impact to their businesses? Could the retail industry have foreseen this sudden surge in demand for hand sanitizers? No retail businesses could have imagined such a major disruption in the supply chain globally.

AI/ML predictions, after all, can only be as good as the data used to generate them.

How is AI/ML being used during the current crisis?

This pandemic has accelerated the use of AI/ML in addressing the immediate needs of varied sectors globally. It is helping health administrations in predicting potential new cases, capacity to provide hospital beds, PPE for doctors and other logistics.

AI mechanisms are being used to track infections, predict hotspots, and keep citizens informed to curb the further spread of the virus. For example, MyGov Corona Helpdesk, a WhatsApp chatbot launched by the Indian government, helps users with verified Covid-19-related information.

Similarly, the Arogya Setu app uses Bluetooth and GPS mechanisms to track infections. Tax, audit and consultancy major KPMG has developed ML models to predict the severity of the outbreak and to identify at-risk populations across the country, state and district level. London-based drug-discovery company BenevolentAI has now turned its attention towards Covid-19.

Industry collaboration on AI/ML

Organizations from different sectors are collaborating to expedite COVID-19 research and development. The Harvard Medical School and Dana Farber Cancer Institute are working with Google Cloud to leverage cloud and analytics technologies to develop a treatment for the virus.

US blockchain hosting provider, Core Scientific, has given Covid-19 researchers access to its AI infrastructure-as-a-service. This service is built on the NetApp Flash-based Storage, Nvidia’s DGX AI systems, and plexus AI stack from Core Scientific. A new consortium of research universities, leading AI company C3.ai and Microsoft - called the C3.ai Digital Transformation Institute - is working on AI techniques to mitigate the implications of the pandemic.

AI/ML in virus research

In the last few years, scientists from around the world have been using sophisticated ML techniques to identify key properties in viruses and flu strains. This can enable scientists to establish commonalities between the strains and choose the right substance for vaccines after examining the protein structure of the virus. A preprint paper published by South Korean company Deargen Inc describes a deep learning model used to predict a drug for the treatment of Covid-19.

AI/ML models incorporating data beyond enterprises’ immediate line of business

For businesses to be better prepared for incidents beyond their control, data arising from their current line of business is simply not sufficient.

According to the British Retail Consortium, weather has the biggest influence on consumer behaviour after the economy. Businesses stock up goods based on changes in weather conditions or seasons. If AI/ML models of businesses integrate data from the weather department along with their routine data, they will be able to predict a short-term surge in demand of emergency supplies in specific areas in a timely manner and the business supply chain can act accordingly.

If the models also understand the demography of a specific area, they will be more accurate in serving consumer needs. IBM Weather Signals uses Watson AI to predict impact in business performance. It does this by merging weather data with a company’s operational data used in a model that can predict how weather conditions like temperature, wind chill, and humidity can impact sales of individual product categories in specific areas.

Similarly, AI models can use Natural Language Processing (NLP) and analyze social media content. This can help businesses understand consumer behaviours, detect trends, and understand target audiences.

If AI models are used on data from hospitals and health departments across the world, they can better predict and understand general consumer and medical supply needs across different parts of the world.

AI/ML platforms using data from diverse sources are emerging

Big businesses could build comprehensive predictive models using a wide variety of data, but many mid to small businesses will need to rely on external sources.

Canadian company BlueDot’s model uses natural language processing and Machine Learning on the data it collects from varied sources. These include statements from official public health organizations, digital media, global airline ticketing data, livestock health reports, population demographics, climate data from satellites, and local information from journalists and healthcare workers.

Regular alerts are sent to businesses, governments and health care clients. They picked up the Covid-19 phenomena nine days before the WHO warned people about the emergence of this virus and have made successful predictions in the past with regards to the Zika virus as well.

The field of AI is changing rapidly across multiple dimensions. The current crisis has only accelerated its adoption. Businesses, research institutes, health departments and governments are pivoting towards the use of AI with many creating shared platforms and investing in joint efforts. Data from numerous sources is being leveraged to make better predictions and more accurate analysis. All these efforts are likely to prepare the world to be more proactive and respond better to a similar crisis in the future.

• Vivek Prakash
• The author is Director, SW Engineering, NetApp