Predictive modeling and data curation guides state in COVID-19 response

By: Denise Blough, College of Public Health

With a dissertation on building models for influenza pandemics and years of experience working with health departments on complex public health problems, Ayaz Hyder was about as prepared as you could be to meet the challenges brought by the COVID-19 pandemic.

Thanks to support from Ohio State’s Office of Research and Office of Outreach and Engagement, the assistant professor of environmental health sciences is using his expertise in computational epidemiology to develop a shared data resource for agencies and organizations that are responding to COVID-19 and leading recovery efforts across Ohio.

Hyder said data sharing is increasingly important to making sense of the dozens of disconnected COVID-19 models that are created and updated daily — predicting the spread of infections, economic impacts, unemployment claims and other pandemic-related developments.

The resource he and his team have built, called COVIDCommons, will bridge the gap between varying types of COVID-19 models and datasets by compiling them in one place.

“COVIDCommons will bring these different models together in a way that allows different types of researchers across Ohio State and beyond to be able to share their data and contribute to a set of tools that can provide information to decision makers,” he said.

The modeling data, to be publicly available, can be used to run ‘What if?’ scenarios for closing and opening different schools and workplaces, as well as predict the impact of various levels of compliance with public health orders surrounding social distancing.

“We can run experiments where we infect certain individuals and see how it affects the trajectory of the epidemic, which is just not possible in the real world,” Hyder said. “It gives policymakers and decision makers a tool to use to figure out what might work best.”

For weeks, Hyder has been using similar methods to run simulation models around reopening Ohio, which has assisted the Ohio Department of Health and other government and community leaders as they make policy decisions. His intricate models use more than 100 parameters based on real-time data sets and prior studies, such as COVID-19 cases in Franklin County, mobility data, disease transmission rates and data on businesses, daycares and schools.

“It’s kind of like the game SimCity, where you essentially create your own artificial society,” Hyder said. “These simulation models are basically a computational replica of the state of Ohio or of Franklin County, and we model the spread of disease based on where people live, who they come into contact with at home, where they work, where they go to school and who they come into contact with when they’re out in the community.”

Because he was able to adapt prior modeling work from an opioid-related project supported by the National Science Foundation, COVIDCommons was able to get up and running in time to provide immediate assistance to health offices around the state, Hyder said. He was also able to form a rapid team with the help of Can’t Stop Columbus, a volunteer-run tech-based nonprofit that began in response to COVID-19.

And he sees broader public health applications down the road.

“We’re hopeful that the work we’re doing will be directly extendable to the opioid epidemic work and the surveillance work that we’ve been doing, and potentially food insecurity and infant mortality,” he said.

A big challenge will be ensuring the needs of Ohio agencies and organizations are met in a timely manner, he said. He and his team based at Ohio State’s Translational Data Analytics Institute are working with the Ohio Departments of Education, Health, Jobs and Family Services, and Transportation; the governor’s office; Columbus Public Health; Franklin County Public Health; the Mid-Ohio Foodbank and others.

“It’s really all hands on deck, and it’s really putting into practice the fact that we’re all in this together.”

Full article:

Share this page
Suggested Articles
Visualizations show health impact of patients' choices

Different views of personal information helps patients understand probabilities When it comes to conversations with patients, there are many ways for healthcare providers to give information about managing cardiovascular disease....

User stories sought for new academic research platform

Ohio State Libraries is seeking user stories to inform a new research tool it is helping to create as part of the Big Ten Academic Alliance consortium. The Shared BigData Gateway project, led by Indiana...

TDAI hosts (virtual) R programming workshops

TDAI is pleased to announce a multi-part workshop for learning about R.  R is a free and open source programming language used by many for statistical computing.  This series of...

Researchers meet at "Research, Short and Sweet"

“Leveraging Big Data” was the theme of a lively “Research, Short and Sweet” event March 30–the first in a series of quarterly networking events for faculty working in Discovery Theme areas. TDA affiliate Huan...

TDAI co-leading first NSF workshop on translational data science

TDAI and the University of Chicago’s Center for Data Intensive Science are co-chairing an NSF-sponsored Translational Data Science Workshop June 26-27 in Chicago. The invitation-only event is designed to build...