Source: Freepik
Copyright: Freepik
License: Creative Commons Attribution (CC-BY)
New Study Using AI-Powered Analysis Identifies Combinations of Social Barriers Impacting Health Outcomes
GALVESTON, Texas – Most doctors’ visits focus on reviewing medical information such as cholesterol levels and symptoms like a persistent cough to diagnose and treat health conditions. However, discussions rarely touch on nonmedical factors, such as whether a patient has reliable transportation to attend follow-up appointments. Missing critical appointments like radiation therapy to treat a lung tumor due to lack of transport could worsen health outcomes and complicate recovery.
A new study led by researchers at the University of Texas Medical Branch (UTMB) and other partner institutions leverages artificial intelligence (AI) to uncover how such nonmedical factors occur together across patients and their risks for outcomes. Findings published in the Journal of Medical Internet Research, a leading medical informatics journal, offer new insights into how combinations of nonmedical factors provide a foundation for more precisely targeted health care policies.
Nonmedical factors, or social determinants of health, such as financial stability and education access can account for up to 50% of health outcomes.
“While previous studies have analyzed how one or a few non-medical factors impact our health, Americans often face multiple barriers leading to different levels of health risks that are not yet well-understood,” says Dr Suresh Bhavnani, lead author and professor of Biostatistics and Data Science at UTMB. “Using human-centered AI methods, we were able to identify and understand patterns of non-medical factors that were previously unknown.”
The study team included investigators from UTMB, Texas State University, University of Texas Health San Antonio, Houston Methodist, and University of Pittsburgh.
Researchers identified 4 clusters of participants and their nonmedical factors, which were examined and interpreted by a panel of experts. For example, Cluster-1, which included “not employed” and “housing insecurity,” among others, was named Socioeconomic Barriers, and Cluster-4, which included “language barriers” and “lack of health care coverage,” was named Sociocultural Barriers.
The study also analyzed the health risks for each of the clusters. For example, Cluster-1 had a significantly higher risk for depression when compared with Cluster-4. The expert panel suggests that this could be because the combination of barriers such as unemployment and housing insecurity were perceived as being more difficult to overcome and therefore more stressful. Such stress in the long term could lead to a higher risk for depression.
“This analysis was made possible because of the All of Us research program organized by the National Institutes of Health,” says Bhavnani. “The program has collected more than one hundred non-medical factors from a wide range of Americans nationwide, making it one of the largest collections of such critical information.”
The results of this study can inform health care policies aimed at reducing disparities in health outcomes across different populations. For instance, the National Alzheimer’s Project Act 2021 currently highlights the importance of addressing the disproportionately higher dementia risk among Black, Hispanic, and low-income older adults by implementing culturally and linguistically tailored programs.
“But race and income, while critical, are approximations of needs, not the needs themselves,” says Bhavnani. “This study provides a new way to allocate resources based on a combination of specific real-world needs and their risks to health. This approach could make healthcare policies more precise and therefore more efficient in the use of limited resources.”
“Our school strongly supports the multidisciplinary team approach taken in this study which included experts in AI, biostatistics, programming, clinical care, health services research, gerontology, and ethics. Such teams are critical for addressing the complex public health challenges faced by our underserved communities,” says Dr Peek, senior vice president and dean, School of Public and Population Health at UTMB.
“This study proposes a novel approach of how AI results related to non-medical factors could be translated into more precise public health policies, an increasingly important goal for translational science,” says Dr Urban, director, Institute for Translational Sciences at UTMB.
This research was funded by the Artificial Intelligence/Machine Learning Consortium to Advance Health Equity and Researcher Diversity (AIM-AHEAD) program at the National Institutes of Health (NIH), and the UTMB Clinical and Translational Science Award.
Original Article:
Bhavnani SK, Zhang W, Bao D, Raji M, Ajewole V, Hunter R, Kuo YF, Schmidt S, Pappadis MR, Smith E, Bokov A, Reistetter T, Visweswaran S, Downer B
Subtyping Social Determinants of Health in the “All of Us” Program: Network Analysis and Visualization Study
DOI: 10.2196/48775
Journal of Medical Internet Research 2025;1:e48775
URL: https://www.jmir.org/2025/1/e48775
About JMIR Publications
JMIR Publications is a leading open access publisher of digital health research and a champion of open science. With a focus on author advocacy and research amplification, JMIR Publications partners with researchers to advance their careers and maximize the impact of their work. As a technology organization with publishing at its core, we provide innovative tools and resources that go beyond traditional publishing, supporting researchers at every step of the dissemination process. Our portfolio features a range of peer-reviewed journals, including the renowned Journal of Medical Internet Research.