Related Experiment Video
Updated: Mar 21, 2026
07:14
Virtual Agent for Real-Time Motivational Interviewing by Integrating Adaptive Nonverbal Behavior and Language Models
Published on: December 23, 2025
591
Effectiveness of Al-Assisted Patient Health Education Using Voice Cloning and ChatGPT: Prospective Randomized
Yan Sun1,2, Shangqing Xu1, Hongying Jin1
1Department of Thoracic Surgery, The First Hospital of Lanzhou University, Donggang West Road 1#, Lanzhou, 730000, China.
Journal of Medical Internet Research
|March 19, 2026
Summary
Patient education using AI voice cloning, especially with the patient's own voice, significantly improves knowledge retention and treatment adherence compared to traditional methods. This AI approach offers a scalable solution for personalized health education.
Area of Science:
- Artificial Intelligence in Healthcare
- Digital Health Interventions
- Patient Education Technology
Background:
- Traditional patient education often lacks personalization, hindering knowledge acquisition and treatment adherence.
- Artificial intelligence (AI), including voice cloning and large language models like ChatGPT, presents opportunities for scalable, interactive health education.
- Limited evidence exists on the comparative effectiveness of AI voice cloning strategies and automated evaluation tools.
Purpose of the Study:
- To evaluate AI-assisted patient education using voice cloning and ChatGPT.
- To compare physician voice cloning versus patient self-voice cloning effectiveness.
- To assess ChatGPT's reliability as an automated evaluation tool for education outcomes.
Main Methods:
- A prospective, 3-arm randomized controlled trial involving 180 hospitalized patients.
- Comparison of traditional education, AI-assisted physician voice cloning, and AI-assisted patient self-voice cloning.
- Primary outcome: education content compliance assessed by ChatGPT-4 and expert review; secondary outcomes: knowledge retention, satisfaction, adherence, quality of life, and psychological status.
Main Results:
- Both AI groups showed significantly higher immediate post-education compliance than the control group (P<.001).
- The patient self-voice group demonstrated superior knowledge retention, satisfaction, and treatment adherence (P≤.02).
- Self-voice group showed sustained adherence and improved psychological well-being and quality of life at 1-month follow-up. ChatGPT evaluations were highly reliable (weighted κ=0.87).
Conclusions:
- AI-assisted patient education, particularly using the patient's own cloned voice, enhances learning outcomes, compliance, satisfaction, and psychological well-being.
- The self-reference effect of using one's own voice improves health education delivery.
- This AI model provides a scalable, cost-effective framework for personalized patient education in clinical settings.
