Using the CRISPE Framework to Assess ChatGPT’s Accuracy of Physical Therapy Examination and Intervention Recommendations for Neck Pain
JOSPT Methods · 2026-07 · 2 (3) · 50-57
DOI: 10.2519/josptmethods.2026.0012 ↗Kurz gefasst
Fachartikel in der Zeitschrift JOSPT Methods, 2026-07. Verfasst von Daniel Peredo und 7 weiteren Autor:innen. Die Arbeit nennt 33 Literaturangaben.
Abstract (Originalsprache)
OBJECTIVE: Examine how manipulating the insight component of the CRISPE (Capacity and Role, Insight, Statement, Personality, and Experiment) prompt framework, affects ChatGPT’s accuracy of physical therapy neck pain recommendations. DESIGN: A 3 × 2 factorial design was used. METHODS: Prompts for GPT-4 were developed using CRISPE with increases in insight on clinical practice guidelines (CPGs) and/or a custom clinical inference table (CIT). Examination (n = 72) and interventions (n = 144) outputs were evaluated by independent teams using precision, recall, and F1 score. Interrater reliability was assessed using intraclass correlation coefficients. Fractional regression models examined effects of conditions with Benjamini-Hochberg adjustment. RESULTS: Fractional regression models showed that attaching the CPG document significantly improved precision for examination (β = 1.30, P < .001) and intervention recommendations (β = 2.74, P < .001). Attaching the CIT improved precision for examination (β = 1.06, P < .001) and intervention (β = 0.98, P < .01). A significant CPG × CIT interaction occurred for recall of examination recommendations (β = 1.39, P < .001). Attaching the CPG improved F1 score for examination (β = 0.38, P < .05) and interventions (β = 2.41, P < .001). Across conditions, F1 scores ranged from 40.2% to 54.5% for examination and from 33.7% to 86.7% for interventions. CONCLUSION: ChatGPT’s accuracy varied by prompt strategy. Attaching guideline resources improved precision and F1 score but not recall. Variability across prompting strategies and accuracy metrics highlights the need for further research. JOSPT Methods 2026;2(3):50-57. Epub 23 April 2026. doi:10.2519/josptmethods.2026.0012
Zitieren
APA
Peredo, D., Paris, E., Ferraro, R., Spano, P., Mohammadi, A., Rao, S., Parrott, J. S., & Vorensky, M. (2026). Using the CRISPE Framework to Assess ChatGPT’s Accuracy of Physical Therapy Examination and Intervention Recommendations for Neck Pain. JOSPT Methods, 2(3), 50-57. https://doi.org/10.2519/josptmethods.2026.0012
Vancouver
Peredo D, Paris E, Ferraro R, Spano P, Mohammadi A, Rao S, et al. Using the CRISPE Framework to Assess ChatGPT’s Accuracy of Physical Therapy Examination and Intervention Recommendations for Neck Pain. JOSPT Methods. 2026;2(3):50-57. doi:10.2519/josptmethods.2026.0012
BibTeX
@article{Peredo2026,
author = {Peredo, Daniel and Paris, Emily and Ferraro, Richard and Spano, Paul and Mohammadi, Asma and Rao, Smita and Parrott, James Scott and Vorensky, Mark},
title = {Using the CRISPE Framework to Assess ChatGPT’s Accuracy of Physical Therapy Examination and Intervention Recommendations for Neck Pain},
journal = {JOSPT Methods},
year = {2026},
volume = {2},
number = {3},
pages = {50--57},
doi = {10.2519/josptmethods.2026.0012}
}
EUVIMED