Journal of Medical Internet Research
Telehealth volume is the primary predictor of artificial intelligence (AI) adoption in U.S. hospitals, according to a study of 6,173 facilities. The research, using 2024 American Hospital Association (AHA) data, found that hospitals with higher telehealth activity are more likely to implement clinical AI tools, though rural facilities face a steeper digital divide.
The study analyzed linked data from the 2024 AHA Annual Survey, the AHA Information Technology Supplement, and the 2023 Healthcare Cost Report Information System (HCRIS). Researchers found that telehealth scale serves as a marker for broader digital readiness, including data infrastructure and organizational capacity.
How does telehealth volume predict AI adoption?
Telehealth volume was the strongest predictor for both clinical and operational AI adoption tiers, according to the study’s multioutput XGBoost models. The association was more pronounced for clinical AI—which includes tools for diagnosis and precision medicine—than for operational AI, such as supply chain or staffing management.
Of the 6,173 hospitals analyzed, 19.2% (1,188) were classified in the highest AI adoption tier (Tier 2). In contrast, 55.7% (3,441) fell into Tier 0, indicating no reported AI adoption. Telehealth volume showed the largest model contribution to high AI adoption across both domains.
Why is there a digital divide in AI readiness?
Geographic data revealed “low-low clusters” of both telehealth and clinical AI adoption across the South, Appalachia, and the Great Plains. Conversely, “high-high clusters” were concentrated in the Northeast and along the West Coast.
The study found that rural hospitals consistently had lower telehealth-associated contributions to predicted clinical AI adoption than metropolitan hospitals, even when telehealth volumes were comparable. This suggests that telehealth expansion alone may not produce equivalent AI readiness in rural settings.
What does telehealth nonreporting signal?
The research distinguished between hospitals that explicitly reported zero telehealth visits and those that failed to report any data. Telehealth nonreporting was heavily concentrated in the lowest AI adoption tier, with 91.4% of Tier 0 hospitals failing to report telehealth volume.
According to the study, this nonreporting may be a signal of broader digital disadvantage rather than a neutral data gap. It may reflect limited administrative reporting infrastructure or a lack of survey response capacity within the facility.
What may happen next for hospital AI?
The study suggests that policies supporting telehealth and AI adoption could be more effective if designed as joint initiatives. Because digital inclusion shapes access to care, focusing on a single technology may be insufficient to close the gap.
Hospitals in rural and structurally disadvantaged communities may require targeted support for interoperability, technical staffing, and governance. A possible next step for these facilities is the redesign of workflows to support AI-enabled care alongside expanded telehealth services.
Frequently Asked Questions
Which AI applications are more strongly linked to telehealth?
Clinical AI applications, such as those used for diagnosis, patient care activities, and clinical decision support, show a stronger association with telehealth volume than operational AI applications.
Do rural hospitals have the same AI readiness as urban hospitals with the same telehealth volume?
No. The study found that at comparable telehealth volumes, rural hospitals showed lower telehealth-associated contributions to predicted clinical AI adoption tiers than metropolitan hospitals.
What data sources were used to determine AI adoption tiers?
The researchers used the 2024 American Hospital Association (AHA) Annual Survey and Information Technology Supplement, linked with 2023 Healthcare Cost Report Information System (HCRIS) data.
Do you think digital infrastructure in rural hospitals should be managed as a public utility to ensure equal access to AI healthcare?