0
0
January 9, 2025
A new study from the University of Minnesota School of Public Health raises critical questions about the use of artificial intelligence (AI) in U.S. hospitals, particularly regarding the accuracy and potential biases of these tools. The research, published in Health Affairs, reveals significant disparities in how hospitals adopt and evaluate AI technologies for daily operations.
The study analyzed data from 2,425 hospitals participating in the 2023 American Hospital Association Annual Survey. It found that 65% of U.S. hospitals use AI-assisted predictive models for tasks like forecasting patient health trajectories, identifying high-risk outpatients, and optimizing scheduling. However, while 61% of hospitals evaluate these tools for accuracy, only 44% assess them for potential bias, raising concerns about inequitable patient care.
“Hospitals with greater financial and technical resources have a distinct advantage,” said Dr. Paige Nong, the study’s lead author and an assistant professor at the University of Minnesota. “They can create models tailored to their patient populations and rigorously evaluate them. In contrast, under-resourced hospitals often rely on ‘off-the-shelf’ solutions that may not align with the unique needs of their communities.”
The findings highlight a growing digital divide in healthcare, with wealthier institutions better positioned to leverage AI for improved outcomes. This disparity could exacerbate existing inequalities in patient care and safety.
The research also points to a lack of standardized procedures for evaluating AI tools, particularly concerning bias. Without consistent evaluation protocols, hospitals risk perpetuating systemic inequities and providing unequal care experiences.
Looking ahead, the research team plans to investigate other AI applications in healthcare, such as ambient scribes and chatbots. They also aim to collaborate with a diverse range of organizations to guide policies and practices that ensure the ethical and equitable use of AI.
“AI has the potential to transform healthcare, but only if implemented thoughtfully and equitably,” Nong emphasized.
For more information, the full study is available in Health Affairs: Paige Nong et al., Current Use And Evaluation Of Artificial Intelligence And Predictive Models In US Hospitals, DOI: 10.1377/hlthaff.2024.00842.
