Data Analytics
How Analytics and Data Management Guided Universities Through the Pandemic
IT and analytics teams have had their time to shine during the past 18 months, as they integrated systems, built dashboards and tackled data governance challenges in order to provide the information needed to keep the campus community safe.
When the pandemic's impact first struck campuses in March 2020, many university leaders developed newfound appreciation for their IT department's data analytics teams, which helped create dashboards offering situational awareness about cases, testing, contact tracing and later vaccination rates. Engagement with analytics accelerated enterprise-wide, as data became critical to health, safety and decision-making.
"IT and analytics have always been valued, but a lot of times we're sort of in the background," said Joe Barnes, chief information security officer at the University of Illinois at Urbana-Champaign (UIUC), who was involved in the university's response efforts. "When the leadership team realized they needed dashboards, and we were able to produce them in short order, they asked us how we did it so fast. It was because we already had the tools. It's been nice to see increased awareness around the value of getting the right data in hand to analyze. I think the leadership definitely has seen the value in terms of how we've been able to gather data and respond."
Scientists at UIUC developed their own COVID test called SHIELD Illinois that requires only a small amount of saliva from an individual and provides results within 24 hours.
It was deployed on the UIUC campus to enable in-person learning during the 2020-21 academic year. Its implementation on a large scale provided quick results, which in turn enabled rapid contact tracing and isolation of positive cases. But buttressing its effectiveness was a data management effort to tie disparate systems together to make testing and tracing work for thousands of students, faculty and staff.
"The analytics team saw that the issue was really just data flow," Barnes explained. Because UIUC's business intelligence teams had already been using Splunk's data platform, they decided to make it a component of the COVID-tracking ecosystem, which they stood up in a matter of weeks. "A student goes to the testing center; their sample is sent to the lab; the lab inputs some information into their system and then that data is sent into Splunk, and then routed over to our health center to do notifications," Barnes said. When the test results would come back, a student with a positive result would get a text within 30 minutes saying someone's going to be reaching out to provide more information. He said Splunk's data platform has been key to the creation of a public data dashboard that has kept students and faculty current with the latest data on testing and new cases.
A software development team at UIUC also built a mobile app to provide pandemic resources to support community health. The Safer Illinois app is used for building access privilege, exposure notification and the option to connect with healthcare professionals.
"We built the Safer Illinois app quickly and the timing was a bit lucky," Barnes recalled. About 18 to 24 months prior to the pandemic starting, the university had begun to create an open source mobile platform geared around student success. "We had our soft launch the fall before the pandemic started," he said. In 2020 they quickly repurposed it as a COVID safety app. "You could use it at a testing center to scan in with the app. When your results came back, you would get your results and test history."
The app also was used for building access. "We hired a bunch of wellness support people to check the apps at building entrances," Barnes said. "You could show your digital ID, and it would say if you were clear to enter or if you were positive or out of testing compliance." The app was regularly updated as the CDC guidelines were changing, and it added vaccine status when the vaccine cards came out. "We had the ability for you to upload your vaccine card into our medical record system, and then that ties back into the Safer app," Barnes said. "We're still using it today for building access."
Data Governance Lessons
Throughout the pandemic, Carnegie Mellon University in Pittsburgh leaned heavily on its analytics teams to pull together multiple data sets to make sense of the situation on campus.
"My team has been core to creating a lot of dashboards and data integrations to making data available to the executive management at the university for making a lot of these decisions," said Roman Mitz, director of enterprise information systems at CMU.
One of the biggest challenges, he added, involved data definitions. For instance, you must define who is a student and who is an active faculty member and exactly where they are and what their status is. "Those are more challenging things to answer at many institutions than you might initially think," Mitz said. "We've actually learned a lot of lessons from the COVID experience to inform our data governance program in terms of how to define the data elements that are necessary. These are long-term issues, but it isn't until you get something like this where you're under a very tight deadline that it really brings things into focus. There are questions we can't answer if we don't have the appropriate data to query it from."
Mitz said that the analytics requirements of the pandemic response definitely put the spotlight on IT. "It dramatically raised our profile in this space on campus and raised the expectations of what we could do. I think it was already trending in that direction. Our CIO is an advocate in this area and I think this is definitely a priority that universities now will want to invest in, if they weren't already."
Metrics for Health and Safety
At Colgate University in upstate New York, a newly formed emergency operations committee set up a health analytics team to gather and help interpret data.
"Our role was to come up with metrics that we felt were important to measure the health and safety of the campus community, as well as the Hamilton community," said Severin Flanigen, financial analyst and health analytics manager. "We're in upstate New York in a very rural setting. We wanted to take that into consideration when we were creating our metrics. Back when the pandemic first started, I think we were tracking some 15 to 20 metrics, each of which had a supporting dashboard that the health analytics team used to determine an alert level. We had a different rubric for each metric that would help inform and guide our conversation as to whether we felt it should be at a moderate or low alert level for each of the different metrics."
When Colgate was preparing to bring students back to campus, the question was how to keep the students safe and use the data to help inform decision-making. "We had a series of 'gate levels,' with gate level zero being an initial two-week quarantine," Flanigen explained. "As they progress through the different levels, the students are granted a little bit more freedom and flexibility to get together and have a more normal college experience. The data we gathered helped inform the current gate level. As we were going through the semester, we would have weekly meetings with the task force on reopening, and if we felt that the data warranted a gate-level change in one direction or another, we could make a recommendation to the task force and provide them our rationale for such a recommendation, which would then go up to the executive group, who would make the final decision."
"At the beginning of the pandemic, we had all these metrics we wanted to create, and we were scrambling to create new processes on the fly," said Tim Borfitz, Colgate's director of data analytics and decision support. "We took advantage of a lot of existing tools, as well as some custom development work. We used some Google Forms; we used our Qualtrics enterprise license to survey and gather information. For instance, we wanted to de-densify the campus as students were returning, so they had to pick a time slot. There was a quota for each time slot within Qualtrics. In addition, members of the data analytics team brought a lot of the data into the data warehouse, so we could join it and do aggregation on test results, and PPE. And we made extensive use of Tableau for dashboards."
The health analytics team got positive feedback from the campus community for its efforts. "Once they saw what we can do with the data, both on the front end and the back end, requests for looking at data and analyzing data is at an all-time high," Borfitz said. "It has grown tremendously."
