Design and implementation of an end-to-end AI-driven colonoscopy recall workflow at scale
JAMIA Open · 05.05.2026 · 9 (3)
DOI: 10.1093/jamiaopen/ooag070 ↗Kurz gefasst
Fachartikel in der Zeitschrift JAMIA Open, online erschienen am 06.05.2026. Verfasst von Aman Mohapatra und 15 weiteren Autor:innen. Beteiligte Einrichtung: Department of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, 02215, United States. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 15 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Beth Israel Lahey Health.
Abstract (Originalsprache)
Objectives
To develop a large-language-model (LLM)–centric workflow flow extraction and migration of clinician-documented colonoscopy recall recommendations from unstructured reports and letters during an enterprise-wide electronic health record (EHR) transition. Materials and Methods A multi-stage workflow [Optical Character Recognition (OCR) -> LLM -> structured fields] was built around a central GPT-4 Turbo inference step following prompt optimization. Validation was performed on a held-out set (N = 326 notes) using 2-clinician consensus and then benchmarked against traditional rule-based natural-language-processing (NLP) (spaCy v3). Layered quality control—manual review, field validation, and anomaly detection—was used to assess workflow results prior to upload (N = 118 181 total patients).
Results
Prompt optimization enabled GPT-4 Turbo to achieve perfect concordance with clinician review in a small test set (macro-F1 = 1.0; N = 100 patients). Expanded validation on a held-out set demonstrated improved F1 (0.89; CI = [0.65, 0.92], N = 326) relative to a traditional rule-based NLP approach (F1 = 0.78; CI = [0.58, 0.82]). The system processed 118 181 records in ≈9 hours (≈2 s/record) at a direct implementation cost of ∼$12 000.
Discussion
An LLM-driven workflow safely migrated preventive-care data at population scale, with potential accuracy improvements over traditional rule-based NLP approaches and substantial reductions in time and cost relative to manual review.
Conclusion
LLMs can play a valuable role in high-quality structuring of clinical data, preserving longitudinal care continuity during EHR modernization.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, 02215, United States.
- Information Systems, Beth Israel Lahey Health, Boston, MA, 02215, United States.
- Halo Solutions, LLC, Boston, MA, 02215, United States.
ORCID-Profile: Aman Mohapatra
Förderung
- Beth Israel Lahey Health
Zitieren
APA
Mohapatra, A., Porth, R., Wong, S., Hardway, H., Piatkowski, G., Shang, J., Amat, M. J., Flier, S., Salsman, A., Fitzgerald, T., Shammout, A., Rubins, D., Miller, A., Jegadeesan, V., Ravi, A., & Feuerstein, J. D. (2026). Design and implementation of an end-to-end AI-driven colonoscopy recall workflow at scale. JAMIA Open, 9(3). https://doi.org/10.1093/jamiaopen/ooag070
Vancouver
Mohapatra A, Porth R, Wong S, Hardway H, Piatkowski G, Shang J, et al. Design and implementation of an end-to-end AI-driven colonoscopy recall workflow at scale. JAMIA Open. 2026;9(3). doi:10.1093/jamiaopen/ooag070
BibTeX
@article{Mohapatra2026,
author = {Mohapatra, Aman and Porth, Rachel and Wong, Si and Hardway, Heather and Piatkowski, Gail and Shang, John and Amat, Maelys J and Flier, Sarah and Salsman, Adam and Fitzgerald, Ted and Shammout, Ayad and Rubins, David and Miller, Amy and Jegadeesan, Venkat and Ravi, Arvind and Feuerstein, Joseph D},
title = {Design and implementation of an end-to-end AI-driven colonoscopy recall workflow at scale},
journal = {JAMIA Open},
year = {2026},
volume = {9},
number = {3},
doi = {10.1093/jamiaopen/ooag070}
}
EUVIMED