Predictive AI Adoption and Operational Management in the U.S. Hospitals: National Evidence on Billing, Scheduling, Evaluation, and Governance

Abstract

Artificial intelligence is moving from experimental clinical prediction into the routine administrative and operational infrastructure of hospitals, yet national evidence remains fragmented across adoption, use cases, evaluation, and governance. This quantitative secondary-data study examines predictive artificial intelligence in United States nonfederal acute-care hospitals using weighted national estimates from the 2023 and 2024 American Hospital Association Information Technology Supplement and peer-reviewed multivariable results from the 2023 survey. The present study asks whether adoption increased, whether operational applications expanded, and whether organizational and technology characteristics were associated with evaluation capacity. Predictive artificial intelligence adoption increased from 66 percent in 2023 to 71 percent in 2024. Among hospitals using predictive models, billing simplification increased from 36 percent to 61 percent and scheduling facilitation from 51 percent to 67 percent. Large, urban, system-affiliated, non-critical-access hospitals and hospitals using the market-leading electronic health record vendor reported substantially higher adoption. In 2024, system members reported higher rates than independent hospitals for accuracy evaluation, bias evaluation, and post-implementation monitoring. Published adjusted analysis of the 2023 data further showed that self-developed models were associated with greater local evaluation, whereas third-party models were associated with lower local evaluation. Bias evaluation remained less prevalent than accuracy evaluation. The evidence is consistent with a Technology-Organization-Environment interpretation in which digital infrastructure, organizational scale, system affiliation, and model sourcing shape the organizational capacity to deploy and govern predictive tools. The present study contributes a nationally grounded view of hospital artificial intelligence management while avoiding unsupported claims about cost, throughput, or quality outcomes that the public data do not measure. The results identify operational diffusion alongside a persistent organizational digital divide in evaluation and governance capability.