Eight researchers at the University of Michigan School of Information have earned grants from the 2024 Michigan Institute for Data and AI in Society (MIDAS) Propelling Original Data Science (PODS) program. 

Their projects will spur innovations in healthcare, environmental sustainability and education through the use of data science and artificial intelligence. The PODS tracks have been organized in three categories: data science and AI methodology and applications, accelerating responsible AI research ecosystems and AI for health policy and healthcare. 

UMSI awardees and their tracks: 

PODS Track 1: Data science and AI methodology and applications

Multimodal Modeling of Cognitive Load at Individual and Team Levels in Acute Care Teams using VR Simulations

Vitaliy Popov, Michael Cole and James Cooke (U-M Medical School), and Mohamed Abouelenien (College of Engineering and Computer Science, U-M Dearborn)

This project studies how critical care teams handle stress during complex healthcare emergency situations. The group’s research will help inform better training methods for healthcare professionals and other high-risk industries, potentially resulting in improved performance and patient care. 

“Our research team is grateful for MIDAS support, which will help us open the ‘black box’ of team cognition in acute scenarios’” Popov says. “This study will be the first to examine the relationship between cognitive load dynamics and behavioral responses at both individual and team levels using objective and self-report data.”

Distributing Expert Attention in Complementary Systems

Sabina Tomkins and Grant Schoenebeck (School of Information), Derek Van Berkel (U-M SEAS), and Ariel Hasell and John Ryan (U-M LSA)

This project aims to answer the question of where to best distribute expert attention between human and AI annotators. 

“Given recent developments in AI there is growing evidence that automated systems can achieve better performance than humans at creating training data for machine learning pipelines. However, it is likely that these AI annotators will still require expert supervision,” Tomkins says. “Our task is to understand how to effectively allocate expert attention when tasks are distributed across humans and AI annotators. We will implement our approach in two high-stakes environmental problem settings: wildlife conservation and climate change.” 

PODS Track 2: Accelerating responsible AI research ecosystems

Innovating, Applying, and Educating on Fairness and Bias Methods for Educational Predictive Models

Christopher Brooks, Libby Hemphill and Allyson Flaster (U-M Institute for Social Research)

UMSI associate professor Christopher Brooks says this project will look at the use of AI models in higher education. These models, typically used by academic advisors, help predict a student’s success in college. Sometimes, Brooks says, these models can make faulty and biased assumptions about students.

“We have what's called true positives and false positives,” Brooks said. “The model may predict people will drop out who actually aren't in danger of dropping out. And that's important because the institution may waste resources on those students instead of students who may need additional help.” 

The grant, Brooks says, will help expand the data pool. Right now, Brooks, Hemphill and Flaster have been looking at data from four institutions. The money from the grant will help them expand to nine institutions over a 15 year period. 

“I think it’s great that MIDAS is investing in this research,” Brooks says. “I think it can have a huge impact on education and how we determine the likelihood of student success in the future.” 

Evaluating GenAI and Team-based Solutions to Reverse the Decline of Online Knowledge Communities

Yan Chen and Qiaozhu Mei

Advancing Responsible AI by Rethinking the Roles of Marginalized Communities in the Innovation Lifecycle: Developing the UBEC Approach

Shobita Parthasarathy and Molly Kleinman (U-M Gerald R. Ford School of Public Policy), Ben Green 

RELATED

Learn more about the Michigan Institute for Data and AI in Society by visiting their website.

Read “MIDAS announces 2024 PODS and DATA awardees” at the University Record. 

— Noor Hindi, UMSI public relations specialist