IBM teams with Rolls-Royce to help Europe respond more effectively to COVID-19 impacted communities
Rolls-Royce and IBM® support the not-for-profit Emergent Alliance, using IBM Cloud Pak® for Data to help accelerate responses to COVID-19 outbreaks.
Collaboration and co-creation are among the driving forces that can produce positive outcomes where they matter most. Now more than ever, decision making relies on data — and organizations must formulate proactive plans to journey forward in challenging times. IBM, Rolls Royce and dozens of global enterprises founded Emergent Alliance, a not-for-profit alliance designed to support innovation and resilience as organizations move toward a post-COVID-19 future.
IBM helped create Emergent Alliance when Andrew Brown, General Manager of IBM Cloud and Cognitive Software Europe and sponsor for Emergent Alliance urged IBM employees to join the mission. The effort was led by Stephen Warwick, Head of IBM UK Labs, Susara van den Heever, IBM Director of Data and AI Expert Labs and Learning in Europe, Middle East and Africa, and Seth Dobrin, Chief Data Officer for IBM Cloud and Cognitive Software, joined by committed supporters from the IBM Data Science and AI Elite team (DSE) who joined the task force to contribute to the effort.
Caroline Gorski, Group Director at Rolls Royce and co-founder of the Emergent Alliance, provided crucial leadership for the project, driving collaborative efforts between the DSE and a passionate group of data scientists from the R2 Data Labs led by Klaus Paul, Berlin AI Hub and Emerging Technologies Capabilities Lead.
The team’s chemistry was immediate. The group agreed that identifying the status of cases in localized areas was crucial to help local authorities mount a more effective response to COVID-19 outbreaks. Starting with a virtual data driven workshop, the team defined the challenge statement: Create a risk-pulse index to help governments track the impact of COVID-19 on several aspects of the economy.
Part of IBM’s contribution to this non-for-profit initiative was its unified data and AI platform IBM Cloud Pak for Data, which the team used to facilitate the collaboration between remote teams using these data and AI services:
- Watson® Studio including AutoAI
- Watson Machine Learning
- Cognos® Analytics
- Watson Knowledge Catalog
AI Trust Driven by Visualizations
The core values of openness and trust marked the engagement from its inception. Working together for the greater good and sharing expertise across communities enabled the team to release the code of the implemented models in a dedicated GitHub repository.
Our teams aimed to create AI solutions that truly support end user needs and decision-making processes. Explainable AI (XAI) was imperative because explainable outcomes help users build trust and confidence in the algorithm and emphasize trust in AI. Data visualization powered by Cognos Analytics bridged the three workstreams together, illuminating the logic hidden behind algorithms, transforming the AI-generated outcomes into actionable results and generating enlightening data narratives.
Using Cognos to transform the data scientists’ notebooks developed in Cloud Pak for Data proved effective in communicating the story and results of the team’s work.
The challenge statement served as a starting point to divide the project into three different workstreams:
Emergent Risk Index: A localized risk index incorporating several aspects related to the COVID-19 spread in addition to exogenous variables (density of the population, percentage of the vulnerable population and other issues) with the aim to support local authority countermeasures.
How quickly is the virus spreading in a specific area?
Outcome: The team used open source data to analyze COVID-19 spread and to predict the localized risk index up to six days. This would support early detection of areas that show a sudden increase in the risk index and promote efforts intervene more quickly to keep the local area safer.
Emergent Pulse: This workstream aims to analyze the behavior and sentiment of the population in response to government countermeasures, focusing the attention on UK news articles and behavior toward the hospitality industry.
How do people feel about the lockdown? Are people following the rules?
Outcome: The team created a dashboard to explore news articles to assess how topics and sentiment of the news changed during the pandemic. The team also explored the ways tourism changed during lockdown by analyzing holiday rentals data. As expected, people preferred remote locations away from high density populations, signaling to the hospitality industry the need to adapt to these new behaviors.
Emergent Simulation: the team leveraged the macroeconomic analysis (input–output model) which represents the interdependencies between economic sectors or industries to estimate the impacts of positive or negative economic shocks and analyzing the ripple effects throughout an economy. This workstream developed what-if scenarios through a simulation engine that could help increase understanding of the deep impact of COVID-19 on the economy.
What is the economic impact of Covid-19? When we will recover?
Output: This simulation studies shock propagation and dissipation and applies this knowledge to the shocks caused by COVID-19 to determine how and when sectors can return to normal. The team produced Emergent Economic Engine app that shows how a shock to one sector of the UK economy propagates throughout the national economic network.
- Experienced a complete remote engagement successfully with adaptable tools and methodologies.
- Obtained trusted and explainable solutions that are shared with the community.
- Successfully engaged with a variety of different companies and volunteers with the aim of working together for the greater good.
Great results come from consistency and endurance. The team concluded the first cycle of sprints successfully and have moved to the second phase of the engagement where prototype dashboards illustrated in this article will be tested and used by policymakers to support the decision making processes of local authorities.
Check out this list of articles covering the ongoing project:
- How Many Patients Will Come Through The Doors?
- Impact of COVID-19 on tourism
- Covid-19 News Classification
- Sentiment analysis of newspapers
- Topic modeling of newspapers
- Input-Output Economics and the Impact of Covid-19
- The effective reproduction number – Generalization and Forecasting
- Calculation of the effective reproduction number – Germany
- Applying geospatial knowledge to the COVID-19 assessments
- Code Fest 2020’s Big Data Challenge: Has Covid-19 Switched Our Green Lights Off?
Klaus Paul, Mehrnoosh Vahdat (Project Lead for phase 2), Sarah Boufelja, Astrid Walle, Kyuhwa Lee, Maria Ivanciu, Vincent Nelis, Shri Nishanth Rajendran, Deepak Srinivasan, Álvaro Corrales Cano, Anthony Ayanwale, Kareem Amin, Mara Pometti, Charlie Stanley, Elaine Begley, Scott Couper, Zadia Alden, Erika Agostinelli (Project Lead for phase 1).