# AI Avatar-Delivered Ear Nose Throat (ENT) Induction: A Pilot Feasibility Study of Confidence and Acceptability

## TL;DR

An AI avatar was used to teach junior doctors ENT skills, improving their confidence but not clearly better than traditional methods.

## Contribution

This is the first pilot study to evaluate AI avatars for ENT induction training and trainee confidence.

## Key findings

- AI avatar training significantly increased trainee confidence across all ENT skill domains.
- Most trainees found the AI avatars clear, but only 57% were willing to use AI for further training.
- Most trainees reported no difference in learning or retention compared to traditional teaching.

## Abstract

Background: Junior doctors in otolaryngology (ENT) often start with varied prior experience, making effective induction essential. Artificial intelligence (AI) avatars are a novel method for delivering standardised educational content. This study evaluated whether an AI avatar-delivered ENT induction course could improve trainee confidence and explored participant perceptions.

Methods: A modular online induction course was developed using AI-generated video avatars (HeyGen platform; HeyGen Inc., Santa Clara, CA, USA). Thirty junior doctors at a tertiary hospital completed the course and rated their confidence in seven ENT skills before and after training on a 10-point Likert scale. Post-course surveys assessed clarity, willingness to use AI in the future, and comparisons with traditional teaching.

Results: All 30 participants completed pre- and post-course assessments. Confidence improved significantly across all domains (e.g., identifying normal endoscopic anatomy: 3.3 → 7.6, p<0.001), with large effect sizes. Similar gains were seen in triaging referrals and airway management. The avatars were generally rated clear (mean 7.8/10). Over half (57%) were willing to undertake further AI courses, 30% were unsure, and 13% were unwilling. Most (67%) reported no difference in overall learning compared with traditional methods, while 20% rated AI less effective and 13% reported enhancement. For retention, 70% reported no change, 13% improvement, and 17% decline.

Conclusions: An AI avatar-delivered ENT induction course significantly improved self-reported confidence and was broadly acceptable, though not universally preferred. Most trainees perceived little difference in learning or retention compared with traditional teaching. These findings support AI avatars as a feasible adjunct for induction training, warranting further evaluation with larger, standardised cohorts and objective outcomes.