Hospital Artificial Intelligence/Machine Learning Adoption by Neighborhood Deprivation

Area of Science:

• Health Informatics
• Health Services Research
• Artificial Intelligence in Healthcare

Background:

• Artificial intelligence/machine learning (AI/ML) can potentially reduce health disparities through improved care coordination.
• Empirical evidence on the impact of AI/ML on health equity remains limited.

Purpose of the Study:

• To analyze variations in AI/ML adoption across hospitals based on their characteristics.
• To investigate AI/ML utilization patterns, specifically concerning neighborhood deprivation.

Main Methods:

• Utilized linked datasets from the American Hospital Association (AHA) 2022 Annual Survey and 2023 Information Technology Supplement.
• Integrated data with the 2022 Area Deprivation Index (ADI) to assess neighborhood socioeconomic status.
• Employed state fixed-effect regressions and a decomposition model to analyze AI/ML implementation predictors.

Main Results:

• Hospitals in high deprivation areas (ADI Q4) were less likely to adopt ML/predictive models (coef = -0.10, P = 0.01) and offer fewer AI/ML workforce applications (coef = -0.40, P = 0.01).
• The model explained 79% of the AI/ML adoption variation between high and low deprivation areas.
• Accountable Care Organization affiliation explained 12%-25% of AI/ML utilization differences.

Conclusions:

• Underutilization of AI/ML in disadvantaged and rural areas, particularly for workforce management and EHRs, may prevent these communities from benefiting from AI-enabled healthcare.
• Value-based payment models present an opportunity to strategically support AI integration in healthcare settings.