From Attitudes to Infrastructure: A Multilevel Review of AI Adoption in Healthcare

The development and advancement of AI have been progressed since 1955. Its growth in the healthcare system has resulted in more improved and efficient patient care. For example, a neural network used to diagnose liver diseases caused by the hepatitis virus achieved 97.59% accuracy and exceeded human diagnostic performance in specific contexts. While these advancements have resulted in a rapid adoption rate by hospitals in the 21 st century, a holistic view encompassing both individual and organizational aspects of AI adoption can help healthcare organizations become more informed and take a proactive approach for unexpected events during the adoption process. The current study used the Preferred Reporting Items for Systematic Reviews and Meta-Analysis' (PRISMA) guidelines to review articles applying the Theory of Planned Behavior and the Technology-Organization-Environment framework in medical settings. Results showed that, among individual factors attitude, subjective norm, and perceived behavioral control affect AI adoption. Among technological factors, while perceived usefulness, data quality, data accuracy, and interoperability drive adoption, technological complexity and lack of transparency in AI systems hinder adoption process. Among organizational factors, organizational readiness, leadership support, training availability, and digital infrastructure facilitate adoption; and limited resources, resistance from healthcare professionals, workflow misalignment hinder adoption process. Among environmental factors, government regulations and legal frameworks, external pressures, policy support, and trust in AI systems may drive or hinder adoption. This combination integrates literature by providing an overview of both micro-level psychological factors and macro-level structural factors. Practically, the synthesis of diverse research results can help healthcare organizations, medical AI developers, and policymakers to develop a realistic perspective toward AI adoption.