Selected Research Projects Propose Responsibly-Developed AI Solutions, Improved Health Outcomes
The IC² Institute and Dell Medical School have selected four interdisciplinary research projects for funding. The researchers represent Moody College of Communication; Cockrell School of Engineering; College of Natural Sciences; Dell Medical School; UT Health Houston School of Public Health Austin; Department of Institutional Reporting, Research, Information and Surveys.
The IC² Institute and Dell Medical School announced today that they will be co-funding four research projects that promise significant advancements in the development of artificial intelligence (AI) to improve health outcomes and reduce health disparities.
The four selected projects cut across a variety of health domains: colorectal health; epilepsy; cardiovascular and brain health; skin cancer. Within these domains, researchers will use AI, including machine learning (ML) techniques and natural language processing, to improve diagnostic accuracy and/or predict health outcomes. Additionally, one of the projects aims to integrate social determinants of health — factors like income, education, neighborhood — into health-oriented predictive models.
The use of AI in health care has the potential to revolutionize the patient experience by helping clinicians deliver personalized, predictive, and preventative care. However, much work remains to ensure that AI models are developed in ways that are equitable, algorithms are free of bias, and AI-driven products are effective across diverse patient populations. Each of the selected projects aims to address these issues of ethical and responsible AI development.
Building Trustworthy AI
All research teams will endeavor to reduce bias in their development of AI applications. “Bias can come into play at so many points along the AI development cycle,” said IC² Institute Executive Director S. Craig Watkins, “from problem formulation to data modeling, deployment, and monitoring. We’re asking researchers to consider how bias distributes across this cycle and to propose ways to reduce these instances.”
As they develop AI/ML models, the research teams will be collecting and using data in innovative ways. Much like a human is only as strong as the food it consumes and absorbs, an AI model is only as good as the data it is fed and trained on — the selection and handling of data is of paramount importance, especially in health care.
One of the teams will use unstructured data — clinical notes and recordings of doctor/patient interactions — in their AI modeling. Two of the teams have proposed the creation of synthetic data — AI-generated, sample-based data — to fill in gaps in existing data sets. A major goal of these projects is to figure out ways to mitigate bias in these specific forms of data collection, preprocessing, and use.
Improving Health Outcomes
While the projects promise to advance the ethical application of AI and support the smart use of data, the ultimate goal is to improve patient outcomes. Researchers will be working in clinical settings, in rural and urban locations, as they endeavor to create better diagnostic and predictive tools that benefit their adult and pediatric patients.
“As we build The University of Texas Medical Center, we have an unprecedented opportunity to transform health and health care delivery,” said Claudia F. Lucchinetti, Dean of Dell Medical School and Senior Vice President for Medical Affairs at The University of Texas at Austin. “Foundational to that effort is effectively moving breakthroughs in research from bench to business to bedside. These projects are great examples of AI-driven innovation with the potential to become diagnoses, treatments and cures that improve — and save — people’s lives.”
The funded projects were selected from a competitive field of 39 proposals submitted by UT researchers — representing 13 schools and colleges — earlier this year. The award’s co-funders, IC² and Dell Med, were intentional in seeking interdisciplinary research teams.
“We want to build a broader, diverse community of health AI researchers,” said Greg Pogue, IC² Institute Deputy Executive Director. “When we issued the call for proposals, we wanted to encourage main campus faculty to collaborate with doctors and nurses to ensure that clinical practice informs the research.”
Teams from the selected projects include researchers from Cockrell School of Engineering, Moody College of Communication, College of Natural Sciences, Dell Medical School, UT Health Houston School of Public Health Austin, and Institutional Reporting, Research, Information and Surveys (IRRIS). Each of the four selected project teams will receive up to $150K in funding. Research will begin this fall and will conclude in Fall, 2025.
The IC² Institute and Dell Med are excited about the impact these projects may have on UT and the broader community. Watkins said, “We believe that the UT / AI research community will benefit from seeing researchers develop techniques and procedures that tackle some of the main ethical dilemmas in the design of AI. And the ultimate benefit should spread to healthcare providers and their patients.”
The selected projects are as follows:
- “Enhancing Pediatric Health Equity Using Machine Learning” (John Michael Virostko, M.D.; Augusto Cesar Ferreira De Moraes)
- “Developing an AI-based Diagnostic Tool for Epilepsy in Low-resource Areas” (Kristina Julich, M.D.; Dan Freedman, M.D.; Eunsol Choi)
- “AI and Clinician Collaboration for Unbiased Colorectal Cancer Diagnosis” (Farshid Alambeigi; Sandeep Chinchali; Joga Ivatury, M.D.)
- “Advancing Equitable and Unbiased Health AI for Diverse Populations” (Ruben Rathnasingham; Adewole Adamson, M.D.; DeLawnia Comer-HaGans; Sharon Ricks; Deepak Chetty; John-Paul Clarke, Shiva Jaganathan)