An automated information extraction model for unstructured discharge letters using large language models and GPT-4

The administrative burden of manually extracting clinical information from discharge letters is a common challenge in healthcare. This study aims to explore the use of Large Language Models (LLMs), specifically Generative Pretrained Transformer 4 (GPT-4) by OpenAI, for automated extraction of diagno...

Full description

Saved in:
Bibliographic Details
Main Authors: Robert M. Siepmann, Giulia Baldini, Cynthia S. Schmidt, Daniel Truhn, Gustav Anton Müller-Franzes, Amin Dada, Jens Kleesiek, Felix Nensa, René Hosch
Format: Article
Language:English
Published: Elsevier 2025-06-01
Series:Healthcare Analytics
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2772442524000807
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1832595275302043648
author Robert M. Siepmann
Giulia Baldini
Cynthia S. Schmidt
Daniel Truhn
Gustav Anton Müller-Franzes
Amin Dada
Jens Kleesiek
Felix Nensa
René Hosch
author_facet Robert M. Siepmann
Giulia Baldini
Cynthia S. Schmidt
Daniel Truhn
Gustav Anton Müller-Franzes
Amin Dada
Jens Kleesiek
Felix Nensa
René Hosch
author_sort Robert M. Siepmann
collection DOAJ
description The administrative burden of manually extracting clinical information from discharge letters is a common challenge in healthcare. This study aims to explore the use of Large Language Models (LLMs), specifically Generative Pretrained Transformer 4 (GPT-4) by OpenAI, for automated extraction of diagnoses, medications, and allergies from discharge letters. Data for this study were sourced from two healthcare institutions in Germany, comprising discharge letters for ten patients from each institution. The first experiment is conducted using a standardized prompt for information extraction. However, challenges were encountered, and the prompt was fine-tuned in a second experiment to improve the results. We further tested whether open-source LLMs can achieve similar results. In the first experiment, primary diagnoses were identified with 85% accuracy and secondary diagnoses with 55.8%. Medications and allergies were extracted with 85.9% and 100% accuracy, respectively. The International Classification of Diseases, 10th revision (ICD-10) codes for the identified diagnoses achieved an accuracy of 85% for primary diagnoses and 60.7% for secondary diagnoses. Anatomical Therapeutic Chemical (ATC) codes were identified with an accuracy of 78.8%. On the other hand, open-source LLMs did not provide similar levels of accuracy and could not consistently fill the template. With prompt fine-tuning in the second experiment, the primary diagnoses, secondary diagnoses, and medications could be predicted with 95%, 88.9%, and 92.2% accuracy, respectively. GPT-4 shows excellent potential for automated extraction of crucial diagnostic and medication information from discharge letters, presumably lowering the administrative burden for healthcare professionals and improving patient outcomes.
format Article
id doaj-art-ac1d469139464923b279eea96d11db05
institution Kabale University
issn 2772-4425
language English
publishDate 2025-06-01
publisher Elsevier
record_format Article
series Healthcare Analytics
spelling doaj-art-ac1d469139464923b279eea96d11db052025-01-19T06:26:56ZengElsevierHealthcare Analytics2772-44252025-06-017100378An automated information extraction model for unstructured discharge letters using large language models and GPT-4Robert M. Siepmann0Giulia Baldini1Cynthia S. Schmidt2Daniel Truhn3Gustav Anton Müller-Franzes4Amin Dada5Jens Kleesiek6Felix Nensa7René Hosch8Department of Diagnostic and Interventional Radiology, University Hospital Aachen, Aachen, GermanyInstitute of Diagnostic and Interventional Radiology and Neuroradiology, University Hospital Essen, Essen, Germany; Institute for Artificial Intelligence in Medicine (IKIM), University Hospital Essen, Essen, GermanyInstitute for Artificial Intelligence in Medicine (IKIM), University Hospital Essen, Essen, Germany; Institute for Transfusion Medicine, University Hospital Essen, Essen, GermanyDepartment of Diagnostic and Interventional Radiology, University Hospital Aachen, Aachen, GermanyDepartment of Diagnostic and Interventional Radiology, University Hospital Aachen, Aachen, GermanyInstitute for Artificial Intelligence in Medicine (IKIM), University Hospital Essen, Essen, GermanyInstitute for Artificial Intelligence in Medicine (IKIM), University Hospital Essen, Essen, GermanyInstitute of Diagnostic and Interventional Radiology and Neuroradiology, University Hospital Essen, Essen, Germany; Institute for Artificial Intelligence in Medicine (IKIM), University Hospital Essen, Essen, GermanyInstitute of Diagnostic and Interventional Radiology and Neuroradiology, University Hospital Essen, Essen, Germany; Institute for Artificial Intelligence in Medicine (IKIM), University Hospital Essen, Essen, Germany; Corresponding author. Institute of Diagnostic and Interventional Radiology and Neuroradiology, University Hospital Essen, Hufelandstraße 55, 45147, Essen, Germany.The administrative burden of manually extracting clinical information from discharge letters is a common challenge in healthcare. This study aims to explore the use of Large Language Models (LLMs), specifically Generative Pretrained Transformer 4 (GPT-4) by OpenAI, for automated extraction of diagnoses, medications, and allergies from discharge letters. Data for this study were sourced from two healthcare institutions in Germany, comprising discharge letters for ten patients from each institution. The first experiment is conducted using a standardized prompt for information extraction. However, challenges were encountered, and the prompt was fine-tuned in a second experiment to improve the results. We further tested whether open-source LLMs can achieve similar results. In the first experiment, primary diagnoses were identified with 85% accuracy and secondary diagnoses with 55.8%. Medications and allergies were extracted with 85.9% and 100% accuracy, respectively. The International Classification of Diseases, 10th revision (ICD-10) codes for the identified diagnoses achieved an accuracy of 85% for primary diagnoses and 60.7% for secondary diagnoses. Anatomical Therapeutic Chemical (ATC) codes were identified with an accuracy of 78.8%. On the other hand, open-source LLMs did not provide similar levels of accuracy and could not consistently fill the template. With prompt fine-tuning in the second experiment, the primary diagnoses, secondary diagnoses, and medications could be predicted with 95%, 88.9%, and 92.2% accuracy, respectively. GPT-4 shows excellent potential for automated extraction of crucial diagnostic and medication information from discharge letters, presumably lowering the administrative burden for healthcare professionals and improving patient outcomes.http://www.sciencedirect.com/science/article/pii/S2772442524000807Large language modelsAutomated information extractionArtificial intelligenceGenerative pre-trained transformer (GPT)ChatGPTDischarge letters
spellingShingle Robert M. Siepmann
Giulia Baldini
Cynthia S. Schmidt
Daniel Truhn
Gustav Anton Müller-Franzes
Amin Dada
Jens Kleesiek
Felix Nensa
René Hosch
An automated information extraction model for unstructured discharge letters using large language models and GPT-4
Healthcare Analytics
Large language models
Automated information extraction
Artificial intelligence
Generative pre-trained transformer (GPT)
ChatGPT
Discharge letters
title An automated information extraction model for unstructured discharge letters using large language models and GPT-4
title_full An automated information extraction model for unstructured discharge letters using large language models and GPT-4
title_fullStr An automated information extraction model for unstructured discharge letters using large language models and GPT-4
title_full_unstemmed An automated information extraction model for unstructured discharge letters using large language models and GPT-4
title_short An automated information extraction model for unstructured discharge letters using large language models and GPT-4
title_sort automated information extraction model for unstructured discharge letters using large language models and gpt 4
topic Large language models
Automated information extraction
Artificial intelligence
Generative pre-trained transformer (GPT)
ChatGPT
Discharge letters
url http://www.sciencedirect.com/science/article/pii/S2772442524000807
work_keys_str_mv AT robertmsiepmann anautomatedinformationextractionmodelforunstructureddischargelettersusinglargelanguagemodelsandgpt4
AT giuliabaldini anautomatedinformationextractionmodelforunstructureddischargelettersusinglargelanguagemodelsandgpt4
AT cynthiasschmidt anautomatedinformationextractionmodelforunstructureddischargelettersusinglargelanguagemodelsandgpt4
AT danieltruhn anautomatedinformationextractionmodelforunstructureddischargelettersusinglargelanguagemodelsandgpt4
AT gustavantonmullerfranzes anautomatedinformationextractionmodelforunstructureddischargelettersusinglargelanguagemodelsandgpt4
AT amindada anautomatedinformationextractionmodelforunstructureddischargelettersusinglargelanguagemodelsandgpt4
AT jenskleesiek anautomatedinformationextractionmodelforunstructureddischargelettersusinglargelanguagemodelsandgpt4
AT felixnensa anautomatedinformationextractionmodelforunstructureddischargelettersusinglargelanguagemodelsandgpt4
AT renehosch anautomatedinformationextractionmodelforunstructureddischargelettersusinglargelanguagemodelsandgpt4
AT robertmsiepmann automatedinformationextractionmodelforunstructureddischargelettersusinglargelanguagemodelsandgpt4
AT giuliabaldini automatedinformationextractionmodelforunstructureddischargelettersusinglargelanguagemodelsandgpt4
AT cynthiasschmidt automatedinformationextractionmodelforunstructureddischargelettersusinglargelanguagemodelsandgpt4
AT danieltruhn automatedinformationextractionmodelforunstructureddischargelettersusinglargelanguagemodelsandgpt4
AT gustavantonmullerfranzes automatedinformationextractionmodelforunstructureddischargelettersusinglargelanguagemodelsandgpt4
AT amindada automatedinformationextractionmodelforunstructureddischargelettersusinglargelanguagemodelsandgpt4
AT jenskleesiek automatedinformationextractionmodelforunstructureddischargelettersusinglargelanguagemodelsandgpt4
AT felixnensa automatedinformationextractionmodelforunstructureddischargelettersusinglargelanguagemodelsandgpt4
AT renehosch automatedinformationextractionmodelforunstructureddischargelettersusinglargelanguagemodelsandgpt4