Effectiveness of Al-Assisted Patient Health Education Using Voice Cloning and ChatGPT: Prospective Randomized Controlled Trial

Background: Traditional patient education often lacks personalization and engagement, potentially limiting knowledge acquisition and treatment adherence. Advances in artificial intelligence (AI), including voice cloning technology and large language models (eg, ChatGPT), offer new opportunities to deliver personalized, scalable, interactive health education. However, evidence regarding the comparative effectiveness of different AI-based voice cloning strategies and reliability of automated AI evaluation tools remains limited. Objective: This study aims to evaluate the effectiveness of AI-assisted patient education integrating voice cloning and ChatGPT, compare physician voice cloning with patient self-voice cloning, and assess the reliability of ChatGPT as an automated evaluation tool for education outcomes. Methods: In this prospective, 3-arm, parallel-group randomized controlled trial, 180 hospitalized patients requiring standardized health education were recruited from a tertiary hospital. Inclusion criteria were age ≥18 years, clear diagnosis requiring health education, clear consciousness, and voluntary participation with informed consent. Exclusion criteria were severe hearing impairment, severe cognitive impairment, expected hospitalization <3 days, or prior participation in similar studies. Using a computer-generated random sequence, participants were randomly assigned (1:1:1) to receive traditional education (control), AI-assisted education using physician voice cloning, or AI-assisted education using patient self-voice cloning, each with identical educational content of equal duration. The primary outcome was education content compliance, evaluated using ChatGPT-4 with validated prompts and verified by expert review. Secondary outcomes included knowledge retention, education satisfaction, treatment adherence, quality of life, and psychological status. Outcome assessors and data analysts, but not participants, were blinded to group allocation. Results: Of 180 randomized participants, 174 (96.7%) completed the trial. Both AI-assisted groups had significantly higher mean education content compliance scores immediately posteducation than the control group (physician voice: 86.7, SD 7.3; self-voice: 92.5, SD 6.8; control: 73.2, SD 8.5; <.001). The patient self-voice group showed superior predischarge knowledge retention, higher education satisfaction, and greater treatment adherence than the other groups (all ≤.02). At the 1-month follow-up, the self-voice group maintained improved adherence (Cohen =0.74) and had significantly lower anxiety and depression scores (all ≤.02) and improved SF-36 quality-of-life domains. ChatGPT-based evaluations demonstrated high reliability (weighted κ=0.87, 95% CI 0.82‐0.91) Conclusions: The innovative patient education model integrating AI voice cloning and ChatGPT is distinct from previous studies primarily relying on standard text-to-speech or professionally recorded content. Using patients’ own cloned voices for health education delivery leveraged the self-reference effect to enhance learning outcomes. Compared with research using clinician-narrated content, this study highlights that self-voice education produces superior outcomes across multiple domains including compliance, satisfaction, and psychological well-being. These findings establish a theoretical and practical framework for personalized AI-driven patient education. In real-world clinical settings, this approach offers a scalable, cost-effective solution to enhance patient engagement, particularly valuable in resource-limited environments where individualized education is challenging to deliver. Trial Registration: Chinese Clinical Trial Registry ChiCTR2500101882; https://www.chictr.org.cn/showprojEN.html?proj=268927