California uses open source solutions and the cloud to create a model of models

Governments, like the state government of California, are in the midst of a transition to a new way of delivering vital information, services and programs using technology and the cloud. Government organizations are adopting approaches pioneered in the technology industry including user-centered design, agile development, data science, modular contracting, and the use of modern technology platforms. Many of these governments are using Amazon Web Services (AWS) to respond quickly and scale to handle unprecedented challenges, like COVID-19, and help them quickly make decisions about how to protect their constituents.

Combatting COVID-19

In the earliest days of COVID-19, the Johns Hopkins University (JHU) Infectious Disease Dynamics Group had one of the best computational prediction models that used state or county population numbers along with transportation data to predict the number of people who potentially would be exposed, infected, and/or hospitalized. This model was built as an open source project and could help inform public health leaders’ responses at both the state and county level. The problem was that the model was running on the JHU on-premises infrastructure and therefore was too slow to react to the rapidly changing situations of the pandemic.

AWS worked with JHU, California Department of Public Health staff, and a team of volunteers led by DJ Patil, former U.S. chief data scientist to quickly scale the infrastructure supporting the model. They used AWS to build an architecture that employed high performance computing (HPC) capabilities as well as Amazon Simple Storage Service (Amazon S3), Amazon Elastic Container Registry (Amazon ECR), Amazon Elastic Compute Cloud (Amazon EC2) Auto Scaling, and AWS Batch. With this architecture, the state managed to reduce the time to onboard the model pipeline and generate a full report from one week for one modeling scenario to under 12 hours for hundreds of scenarios.

Building a model of models

In other locations across the US, public health experts at organizations including Stanford University and the Massachusetts Institute of Technology (MIT) developed additional prediction models that provided additional critical insight into COVID-19. However, there was no easy way to compare results across the models. Analysts were pulling multiple models on multiple screens and comparing print outs of different results. Public health officials needed a better way to aggregate multiple models into a single combined view of prediction trend lines to guide their response efforts.

Using user-centered design and an agile approach, the team in California built the California COVID Assessment Tool (CalCAT). CalCAT is an ensemble of models (similar to looking at the various tracks of hurricane forecasts) that gives public health leaders—and now the public—a comprehensive view across multiple models. CalCAT helps leaders answer three major questions: How fast is COVID-19 spreading right now, at a state and county level? What can we expect over the next two to four weeks? And what are the long-term impacts under different scenarios?

In July, California Governor Gavin Newsom announced making CalCAT available to the public and the release of CalCAT as an open source government solution. The California team is now working with other states and other governments to advise them on deploying CalCAT.

“CalCAT is a story of partnership, innovation, and speed. We engaged academic institutions from across the world to draw attention to California. Our goal was to quickly get practical and actionable information into the hands of decision-makers at the state and local levels,” said Michael Wilkening, special advisor to the Governor on innovation and digital services in California. “To date, CalCAT has had over 250,000 users and over half a million page views, with visitors from every US state and 138 countries. The public repository for CalCAT code has been forked 91 times. We continue to see the influence of the tool on the direction of California’s COVID-19 response.”

Open government solutions

Open source tools have become a valuable tool in the fight against COVID-19. Government agencies are using existing open source services to address new mission-critical community needs. For example, the National Health Service (NHS) in the United Kingdom launched a new text message notification service for individuals quarantined at home with suspected coronavirus symptoms using the open source GOV.UK Notify application. GOV.UK Notify has also been implemented in over 1,500 domestic public services, as well as by the Canadian and Australian governments.

As AWS works with government bodies across the globe, we want to make it easier for them to find these open source solutions. Check out AWS Open Government Solutions, launched earlier this year, which features open source code, standards, and practices implemented by governments around the world.