How AI and data models help governments fight Covid-19
A not-for-profit business group including IBM and Rolls-Royce is using AI and data models to help Europe fight Covid-19 and help prepare an economic recovery
As Covid-19 vaccines roll out, getting economies and societies back to normal after the worst of the pandemic has passed will depend on collaboration between industry and the public sector – and harnessing the power of data and technologies, including artificial intelligence (AI).
Now more than ever, decision making in business, government and societies needs to be guided by reliable data rather than gut feeling.
These are among the reasons why IBM, Rolls-Royce, Microsoft and dozens of global companies recently founded Emergent Alliance − a not-for-profit collaboration specialising in data, analytics and technology.
Last April, data scientists and AI experts at IBM (Data Science and AI Elite Team) and Rolls-Royce (R2 Data Labs) joined a team to work on a crucial pandemic-related challenge: how to get a more accurate and up-to-date regional picture of Covid-19 cases so as to help local authorities mount a more effective response to coronavirus outbreaks.
IBM supplied its AI platform − IBM Cloud Pak for Data − and data science resources to help the project’s teams collaborate remotely. The team also used analytics and data management technologies. In the spirit of collaboration, the code behind AI models and insights is freely available to the public via the Emergent Alliance website and code hosting platform GitHub.
To get a holistic view of the impact of the pandemic, we need to understand its implications for health, society and the economy. The project team began by analysing the disease’s impact on health, then analysed how governments reacted, how public behaviour changed and the effect on the economy.
Calculating Covid risk
During the pandemic, many European governments and local authorities have struggled to track spikes in Covid-19 cases in local areas, work out why cases have increased and use the data to respond more quickly and effectively.
The project was divided into three different workstreams. In the first, IBM and Rolls-Royce created a local Covid-19 risk index which included variables such as, population density, and the percentage of the population that was most vulnerable to the virus.
The team used Covid-19 and demographic open-data sources provided by European agencies such as the European Centre for Disease Prevention and Control to analyse the spread of the virus in the UK and continental Europe.
The technology predicted local risk indexes for up to six days, including geographic clusters of the virus
Dashboard displays helped local authorities detect hotspots and regions with a sudden change in the Covid-19 risk index. A quick intervention from the local authority or tougher lockdowns might be needed to prevent a spike in new cases.
IBM data scientist, Erika Agostinelli, stated: “Explainable AI was our focus because explainable outcomes help users build trust and confidence in the algorithm and emphasise trust in AI. Data visualisation transforms the AI-generated outcomes into readable and actionable results.”
The project team also analysed international data, including differences between countries’ lockdowns and international travel restrictions and quarantines.
The human factor
The second project workstream analysed citizens’ behaviour and sentiment in different parts of the UK, including how they were influenced by media reporting and social media content about the pandemic. How do people behave in lockdown? Are people following the rules?
The team also explored community mobility and the ways tourism changed during lockdown by analysing holiday rental data. As expected, people preferred remote locations away from high-density populations, signalling to the hospitality industry the need to adapt to these new behaviours.
Simulating economic shock − and recovery
The team also used a macro-economic data model to understand Covid-19’s full economic impact. It mapped a country’s economy – including interconnected sectors and industries – to simulate likely ripple effects.
Knowledge management and sharing
To share knowledge and ideas the project team built three technology tools: a data repository, a data selection and labelling tool, and a chatbot. The labelling tool produced a library of Covid-19 infection trends, lockdown measures and economic data.
According to Dr. Mehrnoosh Vahdat, an IBM data scientist and squad lead, a data and asset repository that is empowered with standardised metadata enhances accessibility and reusability of developed assets and improves collaboration within data science projects.
The project team are currently building a Covid-19 virtual assistant using IBM’s Watson Assistant to enable users to ask questions such as, “What regions are currently hotspots in the UK?” and “What is the risk of infection in Nottingham?”, and to get answers.
In the project’s next phase, some UK local authorities will start testing the dashboards. Their feedback will be used to improve the technologies, such as how to respond to the Covid-19 infection rate and how the population is reacting to measures to contain the virus.
It has shown that AI can be used for the public good and can handle complex tasks but be easy to understand.
The project is still at an early stage but results thus far have been promising. It has shown that AI can be used for the public good and can handle complex tasks but be easy to understand. It gives an exciting glimpse of a future where the power of credible open data can better inform decision making by corporates, small businesses and nation states – and help get the economy moving.