Patient trust dynamics in ai-based telemedicine: a social psychological systematic review

Abstract

Background: Artificial intelligence (AI)-based telemedicine increasingly supports remote diagnosis, clinical decision support, and technology-mediated health communication. However, patient trust remains theoretically fragmented because prior studies often treat it as a general adoption variable rather than as a social psychological process. This review aimed to synthesize how patient trust in AI-based telemedicine is formed through social cognition, human-AI interaction, and institutional legitimacy. Method: A systematic literature review was conducted following PRISMA 2020. Scopus was the primary database, while PubMed and Google Scholar were used as complementary sources. Searches covered peer-reviewed literature published from 2015 to 2025 using Boolean combinations of trust, social acceptance, social psychology, human-AI interaction, artificial intelligence, telemedicine, telehealth, and virtual care. Two reviewers independently screened records, extracted data using a predefined matrix, and appraised empirical studies using the Mixed Methods Appraisal Tool, while conceptual papers were assessed for relevance, theoretical clarity, and contribution. Heterogeneity was handled through narrative thematic synthesis. Results: Eleven studies/reports were included. Three trust dimensions emerged: individual cognitive appraisal, including explainability, perceived risk, and prior experience; interactional attribution, including AI agency, role clarity, and communication quality; and socio-structural legitimation, including privacy, fairness, bias mitigation, and institutional governance. Human-AI interaction functioned as the experiential channel through which trust was calibrated, while social acceptance operated as a collective condition shaped by norms, professional identity, and legitimacy. Conclusion: Patient trust in AI-based telemedicine is a multilevel social psychological process. Sustainable implementation requires transparent design, privacy assurance, bias governance, and clear alignment between AI systems and clinical roles.