# Predictors of rural hospital closures in the United States: a systematic review and call for AI-driven early warning systems

## TL;DR

This paper reviews factors causing rural hospital closures in the U.S. and argues for AI tools to predict and prevent future closures.

## Contribution

The paper identifies a gap in using AI for predicting rural hospital closures and proposes a framework for AI-driven early warning systems.

## Key findings

- Five main factors contribute to rural hospital closures: financial distress, workforce shortages, policy issues, economic/demographic conditions, and market competition.
- Current studies rely on traditional statistical models and lack real-time AI tools for prediction.
- The proposed Rural Health Control Tower could use multi-source data to monitor and alert about at-risk hospitals.

## Abstract

Rural hospitals in the U.S. are closing at an alarming rate, threatening healthcare access and economic stability in underserved communities. While numerous studies have explored the causes and impacts of these closures, a predictive framework to proactively identify at-risk hospitals early remains underdeveloped.

This systematic review examines the financial, workforce, policy-related, economic, and demographic pressures, as well as market dynamics that contribute to rural hospital closures and evaluates the extent to which current research supports the use of AI-driven early warning systems to predict and prevent future closures.

A systematic literature search was conducted across PubMed, Embase, CINAHL Plus, and Scopus. Studies published between 2013 and 2024 were screened based on predefined inclusion and exclusion criteria. A total of 15 studies were included and analyzed using a narrative synthesis framework following PRISMA guidelines.

The review identified five primary categories contributing to rural hospital closures: financial distress, workforce shortages, unfavorable policy environments, adverse economic and demographic conditions, and market competition. Most studies employed traditional statistical models and used retrospective data to analyze these issues. None of the studies applied ML models to identify these factors. However, there is a significant gap in the availability of real-time tools that can identify contributing factors, predict closures in advance, and guide timely interventions.

Despite the increasing potential of AI, no studies included in this review have applied ML techniques to predict or prevent rural hospital closures. These findings highlight a critical need for developing real-time, AI-driven early warning systems, such as the proposed Rural Health Control Tower (RHCT), to continuously monitor multi-source data, issue dynamic risk alerts, and support proactive decision-making. The predictors and data sources identified in this review offer a foundation for developing such models, which could improve the timeliness of interventions and promote the sustainability of rural healthcare. Future research should prioritize the development of interpretable, equitable, and community-informed AI tools that are accessible to all.

Not applicable

The online version contains supplementary material available at 10.1186/s12913-025-13847-7.