Institute for Equitable Health Data Science Research announces seed grant awardees exploring AI use in healthcare
Projects on antimicrobial nanomaterials, the cardiometabolic risks of antipsychotics and social media outreach to improve cancer trial diversity were selected as the inaugural round of pilot projects from the Institute for Equitable Health Data Science Research.
The institute, an initiative of the Office of the Vice Chancellor for Research, aims to bridge data science and computing research on campus with healthcare research at UIC. Building on the broad scientific strengths and health equity mission of UIC, the institute unites experts from across campus to maximize the potential and minimize the consequences of using AI in medical research and care.
This summer, the institute held its first annual week-long bootcamp to foster interdisciplinary collaboration across UIC. Twenty-five researchers from across UIC and UI Health formed teams to create unique project ideas and research proposals. The top three proposals, led by interdisciplinary teams spanning five different colleges at UIC, were awarded seed grants from the institute.
“It’s a once in a career opportunity to be able to do this,” said William Ackerman, research associate professor in the College of Medicine, and one of the seed-funding awardees. “I've been in research for a long time now—decades—and I have not really had a university that facilitated something like this. So, it was quite an honor to be able to participate.”
Read more about the research projects and teams below.
Fighting drug resistant bacteria with AI-driven designs
Principal investigator: Lisa M. Stabryla, PhD | Assistant Professor, College of Engineering
Bacterial resistance to antibiotics is a major global health threat. Nanomaterials—substances between 1 nanometer and 100 nanometers in size—applied to medical devices and wound dressings could help prevent infections, especially in vulnerable patients, if designed properly.
A project uniting researchers from the Colleges of Engineering, Medicine and Liberal Arts and Sciences combines machine learning with lab experiments to optimize nanomaterial design. The collaboration will identify promising structures that elicit enhanced antibacterial activity without selecting for negative consequences such as bacterial resistance.
The team uses an evolutionary-informed approach to understand the underlying mechanisms of how bacteria adapt to traditional and emerging antibacterial treatments. They then use that knowledge to create materials that could potentially interrupt this process. The research could eventually expand to include antifungal and antiviral solutions.
“We could spend a lot of time and resources iteratively designing and testing a bunch of different nanomaterial structures,” said Stabryla. “But if we could use machine learning to predict biological response or antimicrobial function based on a set of input design structures, we could identify promising candidates or design parameters worthy of testing out on multidrug-resistant strains in vitro.”
In addition to Stabryla, team members include:
- Jida Huang, PhD, College of Engineering
- Kyunghee Han, PhD, College of Liberal Arts & Sciences
- William E. Ackerman IV, MD, College of Medicine
“Something that really stuck with me from the workshop is that it takes one individual to have a good idea, but it takes a team to make it great,” said Stabryla. “This resonated with me because I had been thinking of this pie-in-the-sky idea for a little while, but there's no way I could be executing it alone. Our interdisciplinary team is allowing this idea to come into fruition, and they're genuinely invested in it just as much as I am, and that’s more than anyone can ask for – an enthusiastic team of researchers to work alongside on impactful, high-risk, high-reward ideas.”
Reducing cardiometabolic risks for children prescribed atypical antipsychotics
Principal investigator: Loretta Hsueh, PhD | Assistant Professor of Clinical Psychology, College of Liberal Arts and Sciences
Atypical antipsychotics can increase cardiometabolic risks, including hypertension, high cholesterol and, eventually, type two diabetes and cardiovascular disease. The impact can be greater for children because they are prescribed earlier in life and may have conditions, including psychosis, depression and anxiety, that increase their likelihood of developing cardiometabolic abnormalities. However, it’s estimated that only about half of children receive the necessary screenings for these risks.
Researchers will employ machine learning to identify the factors contributing to non-compliance or successful screening of cardiometabolic risks in children prescribed these medications. The team aims to use the Cosmos data set, a repository of electronic health record data, to identify the contributing factors.
They will consider whether non-compliance with screenings is due to parent refusal, child refusal, lack of physician knowledge or other factors. Additionally, they expect patient sociodemographic information and specific diagnoses may lead to disparities in screening.
"It's a multifaceted issue. It's a really complicated issue,” said Hsueh. “There are plenty of qualitative studies, there are a handful of quantitative studies, but we don't have a comprehensive understanding of what's really going on, and I think that's what the Cosmos data set can help us do.”
Team members include:
- Negar Soheili, PhD, College of Business
- Vanessa Oddo, PhD, MPH, School of Public Health
- Sabrin Rizk, PhD, OTR/L, College of Applied Health Sciences
Once the factors that correspond with noncompliance are identified, the researchers will utilize their diverse backgrounds to develop interventions to help patients get screened, including applying business analytics techniques to the data set.
“There's just a really cool opportunity to apply lots of different fields to this one problem that has potential for a high impact to change the trajectory of a kid's life,” Hsueh said.
Increasing racial diversity in colon cancer clinical trials
Principal investigator: Mohan Zalake | Research Assistant Professor, College of Applied Health Sciences
Currently, only 6-7% of participants in colon cancer clinical trials are Black, even though Black people have the second-highest incidence rate of colon cancer by race in the U.S. This can lead to biased health research and disparities in health outcomes for patients.
In this project, researchers aim to increase the participation of Black adults in colon cancer clinical trials using AI-driven social media strategies.
Team members include:
- Zisu Wang, PhD, College of Business
- Lu Cheng, PhD, College of Engineering
- Janice Krieger, PhD, Community Outreach and Engagement at Mayo Clinic Comprehensive Cancer Center in Florida
The team plans to leverage their multidisciplinary skills and use a three-stage approach. First, they will study existing best practices for reaching certain populations, including work being done by the University of Illinois Cancer Center. They will also collect data from users on Reddit and Facebook who engage with personalized advertisements and posts.
“We really wanted to use social media as a platform so that we can reach people where they are,” Zalake said.
Once they gather and analyze their data, the team will develop a virtual agent that can help communicate with interested participants, answer questions about the clinical trial and support onboarding.
“The whole [institute bootcamp] process enabled us to come together and apply our expertise in a way that has not been applied before,” Zalake said.
Learn more about the Institute for Equitable Health Data Science Research on their website.