Acceptance and Use of Generative AI in Clinical Rehabilitation: An Interdisciplinary Survey of US Rehabilitation Professionals

Abstract

Objective

To examine rehabilitation professionals’ current clinical use and practice patterns of generative artificial intelligence (GenAI) and examine differences between users and nonusers.

Design

Cross-sectional mixed-methods.

Setting

Online survey.

Participants

A total of 408 professionals responded, with 264 complete surveys (64.7% completion) analyzed quantitatively and 311 valid open-ended responses analyzed qualitatively. The participants included occupational therapists, physical therapists, speech language pathologists, clinical exercise physiologists, and athletic trainers in the United States.

Interventions

Not applicable.

Main Outcome Measures

Researcher-designed survey based on the Technology Acceptance Model and Unified Theory of Acceptance and Use of Technology.

Results

Responses were received from all regions of the US, and were primarily occupational therapists (n=136, 51.5%) and women (n=214, 81.1%). Most participants identified as nonusers (n=154, 58.3%), whereas 41.6% (n=101) reported using GenAI some or most of the time. ChatGPT was the most common platform identified by all participants (n=241, 91.3%). GenAI was recognized by rehabilitation professionals as a tool to support patient care and administrative tasks. Users were significantly more likely than nonusers to perceive GenAI as useful in 5 of 7 domains (P<.001) and easy to use in 3 of 6 domains (P<.001). Logistic regression demonstrated that peer influence was the strongest predictor of use (P<.001, odds ratio 6.16, 95% CI, 2.94-12.93).

Conclusions

Perceptions of usefulness, ease of use, and the social influence of peers significantly influence rehabilitation professionals’ adoption and the use of GenAI within clinical practice.