Using the voice of artificial intelligence for physiotherapy student clinical simulation: a mixed methods study

Background Generative artificial intelligence (GenAI) has the potential to enhance the quality of simulation-based learning activities. The aim of this study is to evaluate whether GenAI could provide appropriate feedback on student performance, and to explore students’ perceptions of two simulation tasks.Methods First-year physiotherapy students completed two GenAI-delivered clinical simulation tasks, using ChatGPT (GPT-4o or GPT-5) Advanced Voice Mode, designed to develop communication skills: a clinical handover (Task A) and a patient interview (Task B). GenAI feedback transcripts were assessed against the Clinical Handover Assessment Tool (CHAT; Task A) and a purpose-built rubric (Task B) to determine whether GenAI identified performance deficits. Thirteen students were subsequently interviewed about their perceptions of the tasks.Results Seventy-four students participated across both activities. Task A produced 95 handovers (mean CHAT score 15.3 ± 1.2 SD), with 121 occasions of unsatisfied CHAT domains; ChatGPT identified and gave feedback on 55 (45%) of these. In Task B, 53 interviews were rated using the rubric, with 36 (68%) rated completely appropriate and 17 (32%) mostly appropriate. Inductive analysis of interviews revealed four themes: realism of AI interactions, teaching and learning insights, diverse positive and negative views on AI feedback, and technology perceptions.Conclusion With appropriate task design and attention to the physical environment, voice-based GenAI can be pedagogically tuned to support communication-based simulation activities. The tasks were well received, and GenAI feedback was mostly appropriate but often failed to identify performance deficits.Practice pointsVoice-based generative AI can deliver realistic, engaging clinical simulation tasks for physiotherapy students.Human oversight remains necessary as deficit detection by GenAI was often inaccurate.Students found GenAI a non-judgmental, always-available practice partner, particularly valuable for repetitive and independent practice.Careful task design and a quiet physical environment are important for successful voice-based GenAI simulation delivery.

Authors

Institutions

Publication Details

Journal
European Journal of Physiotherapy
Published
2026-11-13
DOI
https://doi.org/10.1080/21679169.2026.2756302
Primary Topic
Artificial Intelligence in Healthcare and Education
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Using the voice of artificial intelligence for physiotherapy student clinical simulation: a mixed methods study

Mark Overton, Julia Dray, Daniel C. W. Lee, Bruno Tirotti Saragiotto et al.
7 links
European Journal of Physiotherapy
Artificial Intelligence in Healthcare and Education
article

Using the voice of artificial intelligence for physiotherapy student clinical simulation: a mixed methods study

Mark Overton, Julia Dray, Daniel C. W. Lee, Bruno Tirotti Saragiotto, Jack M. Reeves, Bronwyn Hemsley, Simon Buckingham Shum, Lucy Robertson, Daniel White, Joshua W. Pate, Evin Scanlon
article en

Abstract

Background Generative artificial intelligence (GenAI) has the potential to enhance the quality of simulation-based learning activities. The aim of this study is to evaluate whether GenAI could provide appropriate feedback on student performance, and to explore students’ perceptions of two simulation tasks.Methods First-year physiotherapy students completed two GenAI-delivered clinical simulation tasks, using ChatGPT (GPT-4o or GPT-5) Advanced Voice Mode, designed to develop communication skills: a clinical handover (Task A) and a patient interview (Task B). GenAI feedback transcripts were assessed against the Clinical Handover Assessment Tool (CHAT; Task A) and a purpose-built rubric (Task B) to determine whether GenAI identified performance deficits. Thirteen students were subsequently interviewed about their perceptions of the tasks.Results Seventy-four students participated across both activities. Task A produced 95 handovers (mean CHAT score 15.3 ± 1.2 SD), with 121 occasions of unsatisfied CHAT domains; ChatGPT identified and gave feedback on 55 (45%) of these. In Task B, 53 interviews were rated using the rubric, with 36 (68%) rated completely appropriate and 17 (32%) mostly appropriate. Inductive analysis of interviews revealed four themes: realism of AI interactions, teaching and learning insights, diverse positive and negative views on AI feedback, and technology perceptions.Conclusion With appropriate task design and attention to the physical environment, voice-based GenAI can be pedagogically tuned to support communication-based simulation activities. The tasks were well received, and GenAI feedback was mostly appropriate but often failed to identify performance deficits.Practice pointsVoice-based generative AI can deliver realistic, engaging clinical simulation tasks for physiotherapy students.Human oversight remains necessary as deficit detection by GenAI was often inaccurate.Students found GenAI a non-judgmental, always-available practice partner, particularly valuable for repetitive and independent practice.Careful task design and a quiet physical environment are important for successful voice-based GenAI simulation delivery.

European Journal of Physiotherapy
University of Technology Sydney (AU), Universidade Cidade de São Paulo (BR), University of Newcastle Australia (AU)
University of Technology Sydney
Quality education, Good health and well-being
Artificial Intelligence in Healthcare and Education
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon