Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: Mar 31, 2026

A Metadata Extraction Approach for Clinical Case Reports to Enable Advanced Understanding of Biomedical Concepts

A Metadata Extraction Approach for Clinical Case Reports to Enable Advanced Understanding of Biomedical Concepts

Published on: September 20, 2018

Converting unstructured cardiac catheterization and echocardiography reports into structured data using

Fagen Xie¹, Ming-Sum Lee², Wansu Chen¹

¹Department of Research and Evaluation, Kaiser Permanente Southern California Medical Group, Pasadena, CA 91101, United States.

|March 30, 2026

Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Adverse Perinatal Outcomes by Gestational Surrogacy Status among In Vitro Fertilization Pregnancies.

American journal of perinatology·2026

Same author

Does asthma treatment influence COVID-19 severity? A comparative cohort study of SMART vs. Traditional therapy.

European clinical respiratory journal·2026

Same author

Drivers of disparities in asthma exacerbation and healthcare utilization: a quantitative analysis of neighborhood deprivation in Southern California, United States.

Preventive medicine reports·2026

Same author

Research on the impact mechanism of environmental perception of stadium landscapes on sustainable spectatorship willingness from the perspective of embodied cognition: based on the experience model of "body-environment" interaction.

Frontiers in psychology·2026

Same author

Health Care Use and Recurrence Rate in Hemolytic Disease of the Fetus and Newborn: Retrospective Cohort Study.

JMIR pediatrics and parenting·2026

Same author

Validation of electronic health record-based International Society on Thrombosis and Haemostasis bleeding algorithms in a US health system.

Research and practice in thrombosis and haemostasis·2026

Same journal

Evaluating the impact of artificial intelligence scribes on clinical documentation in primary care: a simulation study.

JAMIA open·2026

Same journal

DataAtlas: automatic generation of data dictionaries using large language models.

JAMIA open·2026

Same journal

An examination of the availability and characteristics of social needs data in the electronic health records: a path to social data harmonization and standardization at Johns Hopkins medicine.

JAMIA open·2026

Same journal

Generative artificial intelligence implementation in REDCap.

JAMIA open·2026

Same journal

Improving readability of layperson abstracts and summaries in oncology using task-specific large language model powered tool: results from the BRIDGE-AI 7 study.

JAMIA open·2026

Same journal

Accuracy of administrative data in ascertaining health conditions: a systematic review.

JAMIA open·2026

See all related articles

This summary is machine-generated.

Transformer-based language models (LMs) accurately extract data from cardiac reports locally. This privacy-preserving method enhances data accessibility for research and patient care.

Area of Science:

Artificial Intelligence in Medicine
Natural Language Processing for Healthcare
Clinical Data Extraction

Background:

Echocardiography and cardiac catheterization reports contain vital clinical data on cardiac function and disease.
Extracting this data from unstructured reports is crucial for research and patient care.
Existing methods may raise privacy concerns due to external processing.

Purpose of the Study:

To evaluate the efficacy of locally run, open-source transformer-based language models (LMs) for extracting clinical data from cardiac reports.
To offer a privacy-preserving alternative to external API-based large LMs.
To enhance the accessibility of structured clinical data for research and patient care.

Main Methods:

Fine-tuning two transformer-based LMs (BioclinicalBERT and BART-Large-CNN) locally using a question-answering approach.

Keywords:

BART-Large-CNN BioclinicalBERT cardiac catheterization echocardiography transformer-based language model unstructured data

Related Experiment Videos

Last Updated: Mar 31, 2026

A Metadata Extraction Approach for Clinical Case Reports to Enable Advanced Understanding of Biomedical Concepts

A Metadata Extraction Approach for Clinical Case Reports to Enable Advanced Understanding of Biomedical Concepts

Published on: September 20, 2018

Utilizing a dataset of 3286 echocardiography and 1884 cardiac catheterization reports from electronic health records.

Assessing model performance via accuracy, precision, recall, and F1-score, and evaluating the impact of training set size.

Main Results:

Both LMs demonstrated high performance (>90% accuracy, precision, recall) across report types.
BioclinicalBERT and BART-Large-CNN achieved comparable results, with BART-Large-CNN slightly outperforming on cardiac catheterization reports.
Performance improved with increased training data, plateauing around 1000 reports.

Conclusions:

Locally deployed, fine-tuned transformer-based LMs effectively extract structured data from unstructured cardiac reports.
This automated information extraction supports advancements in clinical research and patient care applications.
The models provide a secure and efficient solution for leveraging valuable clinical information.