Artificial intelligence-mediated clinical communication between providers and patients or caregivers: scoping review and conceptual framework

Artificial intelligence (AI) is increasingly embedded in clinical communication, yet no unified conceptual framework exists to characterize how AI transforms information exchange between healthcare providers and patients or caregivers. We conducted a scoping review of 67 studies, synthesized using Walker and Avant’s concept analysis framework, to identify the defining attributes, antecedents, and consequences of AI-mediated clinical communication. Defining attributes include medical terminology simplification, human oversight and verification, context-adaptive personalization, empathetic tone simulation, intermediary mediation within a triadic provider-AI-patient structure, continuous availability, and accuracy under irreducible risk of error. Antecedents span technological infrastructure, provider burnout and documentation burden, patient digital literacy, and organizational readiness. Consequences include improved patient comprehension, reduced documentation burden, and enhanced patient engagement, alongside risks of hallucination, automation bias, and erosion of therapeutic relationships. These findings establish a conceptual foundation for the design, evaluation, and governance of AI-mediated clinical communication systems in healthcare delivery.