Hospital AI and Robotics Adoption, Access Inequality, and County Mortality: A National Study Across 3,143 U.S. Counties
Johnson, A.; Gefen, D.; Harrison, T.
Show abstract
Hospital artificial intelligence (AI) and robotics are spreading unevenly across the United States, yet national evidence on how these technologies are associated with hospital performance and population health remains limited. This observational study linked the 2024 American Hospital Association Annual Survey, capturing calendar-year 2023 adoption across 6,166 hospitals, to CMS, CDC, and County Health Rankings outcomes for 3,143 U.S. counties. Conditioning on 2019 baseline performance, hospitals using AI for staff scheduling had roughly 4% higher adherence to the SEP-1 sepsis bundle, and hospitals using AI for routine-task automation had approximately 5.1% lower 30-day pneumonia mortality. At the county level, the clearest timing-aligned outcome was contemporaneous hospital-setting mortality (2023): any access to workflow AI hospitals was associated with 25.5 fewer hospital deaths per 100,000 residents under cross-fitted doubly robust estimation (9.9% lower; non-cross-fitted estimate: 16.9 fewer per 100,000). A historical county benchmark (County Health Rankings premature mortality, 2020-2022) showed a concordant pattern, with the cross-fitted estimate corresponding to 975.8 fewer years of potential life lost per 100,000 before age 75 (p < 0.001), though this outcome predates the 2023 adoption measurement and should be interpreted as a lagged benchmark rather than a contemporaneous association. Workflow AI access was also associated with about 206 fewer preventable hospital stays per 100,000 residents (7.2% lower). Robotics showed narrower and more mixed associations. Yet access reflected a substantial digital divide: only 65.8% of Americans lived within 30 minutes of AI-enabled care, leaving about 114.6 million people outside that catchment (AI access Gini = 0.740), and despite a 56% increase in AI-enabled hospitals between 2022 and 2024, distributional inequality persisted. Overall, the evidence is consistent with adjusted associations between workflow AI access and better rescue outcomes where adoption occurs. Significance StatementIn this study of 6,166 hospitals and 3,143 U.S. counties, hospitals using AI for staff scheduling had roughly 4% higher adherence to the recommended sepsis care bundle, and cross-fitted doubly robust estimation associated workflow AI with 25.5 fewer contemporaneous hospital deaths per 100,000 residents (9.9% lower than comparable unexposed counties; non-cross-fitted baseline: 16.9 fewer per 100,000). A concordant historical benchmark (premature mortality, 2020-2022) and a 7.2% reduction in preventable hospital stays reinforce the pattern. Yet about 114.6 million Americans live beyond a 30-minute drive of any hospital using AI. Only 65.8% of Americans lived within 30 minutes of AI-enabled care (AI access Gini = 0.740), and despite a 56% increase in AI-enabled hospitals between 2022 and 2024, distributional inequality did not improve. The contribution is not a blanket claim that AI saves lives; it is a national estimate of health disparities created when hospital technologies diffuse unevenly, suggesting that extending AI-enabled care to underserved areas may represent a high-return intervention.
Matching journals
The top 6 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
- Detecting COVID-19 infection hotspots in England using large-scale self-reported data from a mobile application: a prospective, observational study 90%
- The Impact of School Closure for COVID-19 on the US Healthcare Workforce and the Net Mortality Effects 89%
- COVID-19 mortality and excess mortality among working-age Californians, by occupational sector: March 2020 through November 2021 89%
Similar papers in this journal
- Novel clinical subphenotypes in COVID-19: derivation, validation, prediction, temporal patterns, and interaction with social determinants of health 91%
- Machine Learning for Real-Time Aggregated Prediction of Hospital Admission for Emergency Patients 91%
- Development and assessment of a machine learning tool for predicting emergency admission in Scotland 91%
Similar papers in this journal
- Disentangling the rhythms of human activity in the built environment for airborne transmission risk: a large-scale analysis of mobility data 92%
- Direct and indirect mortality impacts of the COVID-19 pandemic in the US, March 2020-April 2021 91%
- Gendered hiring and attrition on the path to parity for academic faculty 91%
Similar papers in this journal
- Predicting hospital-onset COVID-19 infections using dynamic networks of patient contacts: an observational study 92%
- Understanding COVID-19 trajectories from a nationwide linked electronic health record cohort of 56 million people: phenotypes, severity, waves & vaccination 90%
- Anosmia and other SARS-CoV-2 positive test-associated symptoms, across three national, digital surveillance platforms as the COVID-19 pandemic and response unfolded: an observation study 89%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.