Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

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

Beyond Refills: Strengthening Prediction and Interpretation of Postdischarge Opioid Use After Spine Surgery.

Anesthesia and analgesia·2026
Same author

Unaddressed Biases in a Retrospective Study of Anticoagulant Prophylaxis in Trauma Patients [Letter].

International journal of general medicine·2026
Same author

Interpreting Phase-Specific EEG Signatures After Aortic Surgery: Vulnerability or Physiology?

Journal of neurosurgical anesthesiology·2026
Same author

Comments on "Steady-state regional cerebral oxygen saturation integrates multidimensional periprocedural factors and predicts long-term mortality after transcatheter aortic valve implantation" by Toya et al.

Journal of anesthesia·2026
Same author

Comment on "Effect of Comedy Movies Watched During Transurethral Resection of the Prostate on Vital Signs and Postoperative Pain".

Pain practice : the official journal of World Institute of Pain·2026
Same author

Unequal Local Anesthetic Dosing and Surgical Heterogeneity: Considerations for Interpreting the ESPB-TEA Non-Inferiority Trial [Letter].

Journal of pain research·2026

Related Experiment Video

Updated: Sep 24, 2025

Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
07:15

Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model

Published on: August 16, 2020

7.0K

Predicting Postoperative Mortality With Deep Neural Networks and Natural Language Processing: Model Development and

Pei-Fu Chen1,2, Lichin Chen3, Yow-Kuan Lin1,4

  • 1Graduate Institute of Biomedical Electronics and Bioinformatics, National Taiwan University, Taipei, Taiwan.

JMIR Medical Informatics
|May 10, 2022
PubMed
Summary
This summary is machine-generated.

A novel deep learning model integrating unstructured clinical text significantly improves predictions of 30-day postoperative mortality. This approach enhances patient risk stratification using electronic health records, outperforming existing methods.

Keywords:
anesthesiaanesthesiologistbidirectional encoder representations from transformersdeep learning modeldeep neural networkelectronic health recordmachine learningnatural language processingneural networkpostoperative mortality predictionprediction modelpreoperative medicineunstructured text

More Related Videos

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma
04:09

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma

Published on: October 10, 2018

8.4K
Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
03:14

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

696

Related Experiment Videos

Last Updated: Sep 24, 2025

Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
07:15

Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model

Published on: August 16, 2020

7.0K
Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma
04:09

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma

Published on: October 10, 2018

8.4K
Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
03:14

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

696

Area of Science:

  • Medical Informatics
  • Artificial Intelligence in Medicine
  • Clinical Prediction Models

Background:

  • Machine learning (ML) models show promise in predicting postoperative mortality.
  • Unstructured clinical text, such as preoperative diagnosis and procedure descriptions, contains valuable risk information.
  • Extracting meaningful insights from clinical text for deep learning (DL) models remains a challenge.

Purpose of the Study:

  • To develop a fusion DL model incorporating structured and unstructured data for predicting 30-day postoperative mortality.
  • To evaluate the effectiveness of ML models using preoperative data with and without free clinical text.

Main Methods:

  • Retrospective analysis of electronic health records (EHRs) from 2016-2020 for patients undergoing general or neuraxial anesthesia.
  • Comparison of deep neural network (DNN) models with other algorithms using identical input features.
  • Integration of a DNN model with bidirectional encoder representations from transformers (BERT) to process clinical text.
  • Performance evaluation using area under the receiver operating characteristic curve (AUROC) and area under the precision-recall curve (AUPRC).

Main Results:

  • The study included 121,313 patients, with a 1.29% 30-day postoperative mortality rate.
  • The BERT-DNN model achieved the highest AUROC (0.964) and AUPRC (0.336).
  • BERT-DNN demonstrated significantly higher AUPRC than other models, including DNN and random forest, and significantly higher AUROC than logistic regression and ASAPS.

Conclusions:

  • The BERT-DNN model significantly improves postoperative mortality prediction by incorporating unstructured text from EHRs.
  • This technique enhances the identification of high-risk patients based on surgical descriptions.
  • The model offers a superior approach compared to methods relying solely on structured data or traditional risk scores.