IBM, Rolls-Royce and Emergent Alliance further COVID-19 response with analytics toolsets

By and Álvaro Corrales Cano | 5 minute read | March 2, 2021

The collaboration of the IBM Data Science and AI Elite (DSE) with Rolls-Royce continues to yield significant results for the Emergent Alliance. This not-for-profit alliance supports innovation and resilience as organizations move toward a post-COVID-19 future. The team’s first project involved creating a risk-pulse index to help governments understand the impact of COVID-19 on several aspects of the economy. The IBM and Rolls-Royce team pressed on developing analytics tools to help policymakers better understand the data used to shape strategies in response to COVID-19.

They aimed to provide further insights to public authorities to make more informed policy decisions on the path to recovery. The new toolset designed analyzes and reports a range of health, economic and societal factors.

The team assessed area health risk levels, behavior change and sentiment in the populations using IBM Cloud Pak® for Data. They investigated how governments responded and measured how public reactions changed over time and how these impacted the economy.

They performed predictive and descriptive analytics of open-source datasets. These datasets, compiled from an end-to-end data and AI pipeline, were disseminated through dashboards, web applications and a simulation engine. Users could analyze and quantify the social and economic effects of COVID-19 and estimate the riskiness of a territory.

The Emergent Alliance selection of tools, the code behind AI models, and resulting insights are freely available to the public via tech blogs and GitHub.

Here are some examples of the capabilities available:

Evaluate Health Factors

In several European countries, regional health risk estimates were made possible by training an AI short-term predictions model to forecast the localized risk index up to six days ahead. Spatial-temporal clustering of COVID-19 infections and deaths data helped identify geographic clusters and outliers in the disease spread. An IBM Cognos Analytics dashboard illustrates the results. With this tool, health authorities can see what parts of the country are likely to experience abnormally high or low infection levels. They use the analysis to hone policy decisions to concentrate extra medical resources or tighten lockdown measures.

Figure 1. Dashboard of COVID-19 spatial clusters.

Compare Government Policies and In-Country Restrictions

As COVID-19 spread, governments started to impose a variety of policies to contain the disease. The team analyzed various countries’ clustering of COVID-19 lockdown measures and detected common, meaningful stringency measures across the board. The results, illustrated in the dashboard, give policymakers a global and comparative view over clusters of countries with similar lockdown measures.

Figure 2. Dashboard of clusters of COVID-19 lockdown measures.

Additionally, the team measured country-specific travel restrictions data, developing dashboards that showed the state of international travel restrictions. This tool tracks the restrictions over time and informs citizens about limitations on entry and the possibility to travel to other countries. View the dashboard here.

Figure 3. Dashboard of country-specific travel restrictions.

A travel advisory chatbot using IBM Watson® Assistant provides the general public with information derived from the results of health and travel restriction workstreams. The tool gives users advice for international and domestic travel within the UK, based on health and travel restriction measures.

Figure 4. Chatbot of health and travel restriction.

Analyze Behavioral Patterns

COVID-19-associated health risks and lockdown measures have significantly impacted people’s behavior. The team determined how to isolate the pandemic’s causal impact on the general public’s commuting habits to analyze mobility. The result is a dashboard allowing users to predict the walking mobility for chosen values of lockdown measures such as school or workplace closing. Authorities can use the dashboard to tailor policies intended to drop mobility to specific levels.

Figure 5. Dashboard of mobility estimator.

The news analytics conducted during the team’s first cycle of COVID-19 studies caught the attention of the Nottingham and Nottinghamshire Integrated Care System (ICS)

in England, leading to a similar analysis on social media data supplied by Emergent Alliance. The team investigated the pandemic’s impact on the local population’s mental health by leveraging topic modeling and sentiment analysis techniques.  This means the ICS could track the COVID-19 conversation broken down into topics such as “science,” “politics,” or “distribution,” with their corresponding level of sentiment on a negative to positive scale. The dashboard provided the ICS with extra insights about the pandemic’s impact on the local communities.

Figure 6. Dashboard of social media analysis.

Examine Economic Interventions

The team built economic interventions modeling that estimates shocks’ impact on specific industrial sectors and how they permeate the whole economy. The model incorporates a policy mechanism to limit the impact of the shock. The mechanism, a stimulus, in the form of extra monetary resources, is injected into the economy by governments or other public authorities.

The user can communicate with the model through a novel app, the Emergent Economic Engine. Shock parameters, including the country where it happens, its duration and relative magnitude, are input to the app. Authorities can choose where to inject an economic stimulus to any sector or the economy as a whole. View the app here.

Figure 7. Emergent Economic Engine app.

Cookie-Cutter is a labeling tool launched to help experts generate and label a library of scenarios. The tool allows browsing COVID-19 infection trends and waves, lockdown measures and clusters, mobility and economic data by country to better understand the data and labeling for further analysis. It also features temporal clusters of lockdown measures and a novel way of fitting Gumbel distributions to the health datasets. View the tool here.

Ongoing Impact and Future Work Capabilities

This set of tools and dashboards are now launched on the Emergent Alliance website to help governments make more informed COVID-19 countermeasures through the development of the following outcomes:

  • Valuable insights: development of dashboards to provide insight and enable users to track and explore the results of our work
  • Web applications: developed for simulation of the COVID-19 social and economic impact on the economy
  • Chatbot and labeling tool: developed to facilitate knowledge management and sharing

The Emergent Alliance continues moving forward, testing and collecting feedback from local authorities and volunteers and improving the analytics pipelines.

Learn more about the IBM Data Science and AI Elite.

Explore Industry Accelerators to get your data science project on the fast track to deployment.

Learn more about the Emergent Alliance.

Alliance contributors:
Data Science team: Klaus Paul, Mehrnoosh Vahdat (Phase 2 Project Lead), Álvaro Corrales Cano, Sarah Boufelja, Shri Nishanth Rajendran, Deepak Srinivasan, Elaine Begley, Giorgos Aniftos, Anthony Ayanwale, Ananda Pal, Kyuhwa Lee, Damiaan Zwietering, Amr Abd El Latief, Dirk Ducar, Amritpaul Sohal, Erika Agostinelli (Phase 1 Project Lead), Kareem Amin, Mara Pometti, Astrid Walle, Maria Ivanciu, Vincent Nelis.

Project support: Scott Couper, Zadia Alden, Dean Pellizer, Charlie Stanley, Frank Ketelaars, Claus Samuelsen, Stephen Green.