# An empirical study on the psychological impact of medical AI on patients undergoing dental surgery

## TL;DR

This study explores how AI in dental surgery affects patients' anxiety and satisfaction, finding that trust in AI and transparency play key roles.

## Contribution

The study introduces a trust-augmented framework to show how AI impacts patient anxiety and satisfaction in dental surgery.

## Key findings

- AI significantly reduces patient treatment anxiety and enhances postoperative satisfaction.
- Trust in AI functions as a partial mediator, with effects strengthening in more realistic settings.
- Gender and transparency moderate AI's impact, with higher transparency leading to greater anxiety relief.

## Abstract

With the rapid integration of artificial intelligence technology into the medical field, this study focuses on dental surgery and systematically examines the complex impact mechanisms of AI on patients’ treatment anxiety and postoperative satisfaction through three progressive experiments involving a total of 470 participants. Guided by a TAM-informed, trust-augmented framework, we examine the mediating role of trust in AI and the moderating effects of gender and information transparency. Key findings indicate: (1) AI technology significantly reduces patient treatment anxiety levels (p < 0.001) and enhances postoperative satisfaction (p < 0.001); (2) Across the three studies, trust in AI functioned as a partial mediator overall. The mediation did not emerge in the text-based simulation (Study 1), emerged in the video simulation (Study 2), and became robust in the clinical field study (Study 3; indirect effect 95% CI [− 0.89, − 0.21]) while the direct paths remained significant; (3) Gender differences significantly moderate the effects of AI technology, with female patients showing lower levels of technology trust; (4) Technology transparency positively moderates the therapeutic effect of AI application. Under high transparency conditions, the anxiety relief effect of AI technology (ΔM = 3.04) is significantly stronger than that under low transparency conditions (ΔM = 2.06). Collectively, the findings indicate that trust in AI operates as a partial, context-dependent mediator whose magnitude increases with ecological validity and transparency, clarifying when and how AI use relates to patient anxiety and satisfaction.

The online version contains supplementary material available at 10.1038/s41598-025-29754-0.