people in city wearing masks (petovarga/Shutterstock.com)

Maximizing a vaccine campaign by analyzing social interactions

As decisions are being made about the most effective way to deliver the COVID-19 vaccine to various populations, researchers have developed an algorithm that may optimize its distribution, possibly leading to as much as a seven-fold reduction in the number of infected persons.

Using a synthetic social contact network of residents of Portland, Ore., researchers from Washington State University (WSU) and Pacific Northwest National Laboratory (PNNL) built an algorithm that identifies for vaccination the people that have the maximum reachability into the social network.

“The spreading of a disease is incredibly complex, and the mathematical-based models aren’t well suited to take a network view of people actually interacting with other people,” Ananth Kalyanaraman, WSU’s Boeing Centennial Chair in Computer Science, told WSU Insider. “Our algorithm provides a network view of a population on the move.”

Mathematical models are limited to solving for groups of about 10,000 people, and they can take days to finish their calculations – an impractical solution when millions of people risk infection by their interactions.

WSU researchers had already been collaborating with scientists at PPNL on developing scalable graph applications for influence maximization by identifying influential actors in a network. These tools could, for example, maximize an ad campaign’s influence. When the pandemic hit, they realized their work could also be applied to maximizing the impact of a vaccine campaign. Bringing in partners with expertise in epidemiology from the University of Virginia, the team identified an optimal set of network nodes for vaccination.

“We had a pretty good start,” said Kalyanaraman. “It was a different way of studying a diffusion problem.”

The researchers used the Summit supercomputer at Oak Ridge National Laboratory, currently the world’s fastest high-performance machine.  With Summit’s CPU-GPU nodes, they were able to reduce the time to solution from hours to minutes.

“Speed is a critical factor, and we now have the ability to not just compute the largest number of possible solutions but also compute them quickly and accurately, so that critical problems are addressed in as close to real-time as possible,” said Mahantesh Halappanavar, a computer scientist at PNNL with a joint appointment at WSU.

Still, the algorithm needs significant work before it can be used for an actual vaccine distribution strategy. In particular, the research accounts for individuals in a whole city and their contacts, but it doesn’t include any of the social or economic determinants that also play a role in disease spread.

One bright spot in current COVID‑19 pandemic, according to WSU officials, is the massive amount of data that has been generated about the virus, patients and treatments that will allow scientists to identify intervention strategies for similar challenges in the future.

About the Author

Connect with the GCN staff on Twitter @GCNtech.

Featured

  • business meeting (Monkey Business Images/Shutterstock.com)

    Civic tech volunteers help states with legacy systems

    As COVID-19 exposed vulnerabilities in state and local government IT systems, the newly formed U.S. Digital Response stepped in to help. Its successes offer insight into existing barriers and the future of the civic tech movement.

  • data analytics (Shutterstock.com)

    More visible data helps drive DOD decision-making

    CDOs in the Defense Department are opening up their data to take advantage of artificial intelligence and machine learning tools that help surface insights and improve decision-making.

Stay Connected