Gaps in Artificial Intelligence Research for Rural Health in the United States: A Scoping Review
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
Artificial intelligence (AI) has impacted healthcare at urban and academic medical centers globally. The current focus on AI deployments in urban areas and the history of US urban-rural digital divides raises concerns that the promise of AI may not be realized in rural communities. This may exacerbate well-documented health disparities. Without the benefits of AI-driven improvements in patient outcomes and increased efficiency, rural healthcare facilities may fall farther behind their urban counterparts and rural hospital closure rates may continue to rise.
Methods
We conducted a scoping review following the PRISMA guidelines. We included peer-reviewed, original research studies indexed in PubMed, Embase, and WebOfScience after January 1, 2010 and through April 29, 2025. Studies were required to discuss the development, implementation, or evaluation of AI tools in rural US healthcare, including frameworks that help facilitate AI development (e.g., data warehouses).
Findings
Our search strategy found 26 studies meeting inclusion criteria after full text screening with 14 papers discussing predictive AI models and 12 papers discussing data or research infrastructure. AI models most commonly targeted resource allocation and distribution. Few studies explored model deployment and impact. Half noted the lack of data and analytic resources as a limitation to both development and validation. None of the studies discussed examples of generative AI being trained, evaluated, or deployed in a rural setting.
Interpretation
Practical limitations may be influencing and limiting the types of AI models evaluated in the rural US. We noted validation of tools in the rural US was underwhelming, and ultimately, neglected. With few studies moving beyond AI model design and development stages, there is a clear gap in our understanding of how to reliably validate, deploy, and sustain AI models in rural settings to advance health in all communities.
Funding
National Library of Medicine
Research in context
Evidence before this study: Clinical artificial intelligence (AI)—both for prediction modeling and generative tools— tools promise to reduce care delays, improve diagnosis and treatment decision-making, reduce care costs, and improve efficiency to reduce provider workload and enhance practice management. Unfortunately, efforts to deploy artificial intelligence (AI)—both for prediction modeling and generative tools—in healthcare are advancing, primarily at large academic medical centers and in urban areas. An emerging new digital divide in the use of clinical AI could exacerbate the well-documented health disparities between urban and rural communities in the United States. A better understanding of if and how AI is being developed, deployed, and evaluated across rural US communities is necessary to identify resources gaps and challenges to broad AI use in all communities.
Added value of this study: This study analyzes the current state of artificial intelligence research in the rural United States. For predictive AI models, applications most commonly targeted resource allocation and distribution. We noted several attempts to predict resource utilization of patients who were either tested or tested positive to COVID-19. However, we noted few AI solutions for acute medical events faced by rural patients, such as trauma and stroke, despite worse outcomes for rural patients suffering from these acute events. The limited availability of time-critical specialties such as trauma/emergency medicine, neurology, and cardiology in rural areas often necessitates patients with such conditions be transferred to larger, more resourced hospitals. Practical limitations may be influencing and limiting the types of AI models evaluated in rural US medical facilities. The most frequent model employed were tree-based ensembles, such as random forests and gradient-boosting trees. Our review also highlighted few studies of AI models moving beyond the design and develop stages, leaving a clear gap in our understanding of how to deploy and sustain predictive AI models in rural settings. Several challenges noted in the reviewed studies may provide insight into this lack of translation from research to implementation. We note that validation of A tools in the rural US was underwhelming, and ultimately, neglected. The most common form of model validation employed was a single random holdout test set. Half of the included papers mentioned a lack of reliable data sources or limited data volume as a potential challenge in developing and adopting AI/ML tools. The use of patient-level EHR data was often limited to what was available to the authors or at a specific medical center.
Implications of all the available evidence: Our review indicates a gap and highlights opportunity for innovation in leveraging AI tools to predict and support patients in rural communities. Further research is needed to enhance the translation of state-of-the-art modeling techniques into effective AI tools for use in the rural US, including exploring partnerships between academic medical centers and rural communities and solutions to logistic challenges of such partnerships, including data and resource sharing.