EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Quantifying delay: modeling the impact of timeliness on narrative feedback for Entrustable Professional Activity assessments in internal medicine training using artificial intelligence

Enoch Yu, Haozhong Tian, Farida Mohamad, Karen Schultz, Laura McEwen, Stephen Gauthier, Heather Braund, Nicholas Cofie, Nancy Dalgarno, Adam Szulewski, Farhana Zulkernine, Benjamin YM Kwan

Canadian Medical Education Journal · 30.06.2026

DOI: 10.36834/v1anvj69 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Canadian Medical Education Journal, online erschienen am 30.06.2026. Verfasst von Enoch Yu und 11 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 56 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Introduction

In competency-based medical education (CBME), specifically entrustable professional activities (EPAs), timely feedback is essential because delays can distort assessor recall and performance documentation. Prior work links longer timelag to reduced feedback quality, but its effects across thresholds and outcomes remain unclear. This study examines these relationships using statistical analyses, incorporating artificial intelligence (AI) tools to standardize extraction of narrative feedback metrics.

Methods

We performed a retrospective cross-sectional study of internal medicine resident EPA assessments, each including two narrative sections (‘General Feedback,’ ‘Next Steps’), entrustment scores (0-4), and completion dates. Pretrained transformer-based models estimated sentiment positivity (0-1) and Quality of Assessment for Learning (QuAL) scores (0-5). We manually refined outputs, subsequently fitting datapoints for ordinal logistic, binary logistic, and linear regression. Exploratory analyses identified timelag thresholds linked to changes in feedback metrics.

Results

301 assessors evaluated 1,147 EPAs for 111 residents, with a 7-day median timelag (IQR: 21). Longer timelag was associated with lower QuAL (OR=0.997/day; p < 0.05), fewer words (-0.07words/day; p

Zitieren

APA

Yu, E., Tian, H., Mohamad, F., Schultz, K., McEwen, L., Gauthier, S., Braund, H., Cofie, N., Dalgarno, N., Szulewski, A., Zulkernine, F., & Kwan, B. Y. (2026). Quantifying delay: modeling the impact of timeliness on narrative feedback for Entrustable Professional Activity assessments in internal medicine training using artificial intelligence. Canadian Medical Education Journal. https://doi.org/10.36834/v1anvj69

Vancouver

Yu E, Tian H, Mohamad F, Schultz K, McEwen L, Gauthier S, et al. Quantifying delay: modeling the impact of timeliness on narrative feedback for Entrustable Professional Activity assessments in internal medicine training using artificial intelligence. Canadian Medical Education Journal. 2026. doi:10.36834/v1anvj69

BibTeX
@article{Yu2026,
  author = {Yu, Enoch and Tian, Haozhong and Mohamad, Farida and Schultz, Karen and McEwen, Laura and Gauthier, Stephen and Braund, Heather and Cofie, Nicholas and Dalgarno, Nancy and Szulewski, Adam and Zulkernine, Farhana and Kwan, Benjamin YM},
  title = {Quantifying delay: modeling the impact of timeliness on narrative feedback for Entrustable Professional Activity assessments in internal medicine training using artificial intelligence},
  journal = {Canadian Medical Education Journal},
  year = {2026},
  doi = {10.36834/v1anvj69}
}
Kostenlos starten →