EUVIMEDEUVIMED entdecken
Open Access

Engineering of Generative Artificial Intelligence and Natural Language Processing Models to Accurately Identify Arrhythmia Recurrence

Ruibin Feng iD, Kelly A Brennan iD, Zahra Azizi iD, Jatin Goyal iD, Brototo Deb iD, Hui Ju Chang iD, Prasanth Ganesan iD, Paul Clopton iD, Maxime Pedron, Samuel Ruipérez-Campillo iD, Yaanik B Desai iD, Hugo De Larochellière iD und 5 weitere

Circulation: Arrhythmia and Electrophysiology · 2025

DOI: 10.1161/circep.124.013023 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Circulation: Arrhythmia and Electrophysiology, 2025. Verfasst von Ruibin Feng und 16 weiteren Autor:innen. Beteiligte Einrichtung: Department of Medicine, Stanford University, CA (R.F., K.A.B., Z.A., J.G., B.D., H.J.C., P.G., P.C., M. Pedron, S.R.-C., Y.B.D., H.D.L., T.B., M. V. …. Hauptthemen laut MEDLINE: Natural Language Processing, Arrhythmias, Cardiac, Electronic Health Records, Artificial Intelligence. Die Arbeit wurde bisher 9-mal zitiert (Crossref) und nennt 81 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: NHLBI NIH HHS.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

BACKGROUND: Large language models (LLMs) such as Chat Generative Pre-trained Transformer (ChatGPT) excel at interpreting unstructured data from public sources, yet are limited when responding to queries on private repositories, such as electronic health records (EHRs). We hypothesized that prompt engineering could enhance the accuracy of LLMs for interpreting EHR data without requiring domain knowledge, thus expanding their utility for patients and personalized diagnostics. METHODS: We designed and systematically tested prompt engineering techniques to improve the ability of LLMs to interpret EHRs for nuanced diagnostic questions, referenced to a panel of medical experts. In 490 full-text EHR notes from 125 patients with prior life-threatening heart rhythm disorders, we asked GPT-4-turbo to identify recurrent arrhythmias distinct from prior events and tested 220 563 queries. To provide context, results were compared with rule-based natural language processing and Bidirectional Encoder Representations from Transformer-based language models. Experiments were repeated for 2 additional LLMs. RESULTS: In an independent hold-out set of 389 notes, GPT-4-turbo had a balanced accuracy of 64.3%±4.7% out-of-the-box at baseline. This increased when asking GPT-4-turbo to provide a rationale for its answers, a structured data output, and in-context exemplars, to a balanced accuracy of 91.4%±3.8% ( P

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Natural Language ProcessingArrhythmias, CardiacElectronic Health RecordsArtificial IntelligenceHumansRecurrenceMaleFemalePredictive Value of TestsGenerative Artificial Intelligence

Stichwörter der Autor:innen

artificial intelligenceclinical decision supportelectronic health recordsnatural language processingventricular tachycardia

Autor:innen und Institutionen

  • Department of Medicine, Stanford University, CA (R.F., K.A.B., Z.A., J.G., B.D., H.J.C., P.G., P.C., M. Pedron, S.R.-C., Y.B.D., H.D.L., T.B., M. V. P, M.R., A.J.R., S.M.N.).
  • School of Information, University of California, Berkeley, CA (B.D., S.M.N.).
  • Department of Computer Science, ETH Zurich, Switzerland (S.R.-C.).
  • CoMMLab, Universitat Politècnica de València, Spain (M.R.).

ORCID-Profile: Ruibin Feng, Kelly A Brennan, Zahra Azizi, Jatin Goyal, Brototo Deb, Hui Ju Chang, Prasanth Ganesan, Paul Clopton, Samuel Ruipérez-Campillo, Yaanik B Desai, Hugo De Larochellière, Tina Baykaner, Marco V Perez, Albert J Rogers, Sanjiv M Narayan

Förderung

  • NHLBI NIH HHS

Zitieren

APA

Feng, R., Brennan, K. A., Azizi, Z., Goyal, J., Deb, B., Chang, H. J., Ganesan, P., Clopton, P., Pedron, M., Ruipérez-Campillo, S., Desai, Y. B., De Larochellière, H., Baykaner, T., Perez, M. V., Rodrigo, M., Rogers, A. J., & Narayan, S. M. (2025). Engineering of Generative Artificial Intelligence and Natural Language Processing Models to Accurately Identify Arrhythmia Recurrence. Circulation: Arrhythmia and Electrophysiology. https://doi.org/10.1161/circep.124.013023

Vancouver

Feng R, Brennan KA, Azizi Z, Goyal J, Deb B, Chang HJ, et al. Engineering of Generative Artificial Intelligence and Natural Language Processing Models to Accurately Identify Arrhythmia Recurrence. Circulation: Arrhythmia and Electrophysiology. 2025. doi:10.1161/circep.124.013023

BibTeX
@article{Feng2025,
  author = {Feng, Ruibin and Brennan, Kelly A and Azizi, Zahra and Goyal, Jatin and Deb, Brototo and Chang, Hui Ju and Ganesan, Prasanth and Clopton, Paul and Pedron, Maxime and Ruipérez-Campillo, Samuel and Desai, Yaanik B and De Larochellière, Hugo and Baykaner, Tina and Perez, Marco V and Rodrigo, Miguel and Rogers, Albert J and Narayan, Sanjiv M},
  title = {Engineering of Generative Artificial Intelligence and Natural Language Processing Models to Accurately Identify Arrhythmia Recurrence},
  journal = {Circulation: Arrhythmia and Electrophysiology},
  year = {2025},
  doi = {10.1161/circep.124.013023}
}
Kostenlos starten →