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

A Pilot Study of Ex-vivo High Frequency Ultrasound and Core Needle Biopsy to Improve Clinical Tumor Staging of Cancer of the Bladder.

Urology·2026
Same author

Why machine learning fails at mass spectrometry for small molecules.

Nature metabolism·2026
Same author

Water Vapor Therapy for the Management of Prostate Cancer: The VAPOR 2 Study Initial Results of Ablative Efficacy and Toxicity.

The Journal of urology·2026
Same author

Reply to: Limited evidence of AI superiority in seasonal influenza vaccine strain selection.

Nature medicine·2026
Same author

Prospective deployment of AI-based risk stratification to enable expedited mammography workflow in a safety-net setting.

NPJ digital medicine·2026
Same author

Protein FID: improved evaluation of protein structure generative models.

Bioinformatics (Oxford, England)·2026

Related Experiment Video

Updated: Dec 7, 2025

Detection and Isolation of Cancer in Prostate Biopsies Using Stimulated Raman Histology and Artificial Intelligence
08:05

Detection and Isolation of Cancer in Prostate Biopsies Using Stimulated Raman Histology and Artificial Intelligence

Published on: June 10, 2025

959

Exploiting Rules to Enhance Machine Learning in Extracting Information From Multi-Institutional Prostate Pathology

Enrico Santus1, Tal Schuster1, Amir M Tahmasebi2,3

  • 1Department of Electrical Engineering and Computer Science, CSAIL, MIT, Cambridge, MA.

JCO Clinical Cancer Informatics
|October 2, 2020
PubMed
Summary

Hybrid machine learning (ML) models combining rules and ML algorithms improve clinical note mining performance, even with limited data. These systems outperform traditional rule-based or ML approaches alone.

More Related Videos

Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
05:33

Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System

Published on: July 11, 2025

612
Author Spotlight: Advancing Prostate Cancer Research Through Improved Tissue Sampling and Biobanking
07:34

Author Spotlight: Advancing Prostate Cancer Research Through Improved Tissue Sampling and Biobanking

Published on: November 17, 2023

994

Related Experiment Videos

Last Updated: Dec 7, 2025

Detection and Isolation of Cancer in Prostate Biopsies Using Stimulated Raman Histology and Artificial Intelligence
08:05

Detection and Isolation of Cancer in Prostate Biopsies Using Stimulated Raman Histology and Artificial Intelligence

Published on: June 10, 2025

959
Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
05:33

Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System

Published on: July 11, 2025

612
Author Spotlight: Advancing Prostate Cancer Research Through Improved Tissue Sampling and Biobanking
07:34

Author Spotlight: Advancing Prostate Cancer Research Through Improved Tissue Sampling and Biobanking

Published on: November 17, 2023

994

Area of Science:

  • Medical Informatics
  • Natural Language Processing
  • Machine Learning

Background:

  • Clinical note mining often relies on hand-crafted rules due to limited annotated data, despite machine learning (ML) showing superior performance.
  • Existing ML approaches typically require large training datasets, which are not always feasible in clinical settings.
  • This study addresses the challenge of scarce annotations in clinical note mining.

Purpose of the Study:

  • To develop and evaluate methods for leveraging knowledge from pre-existing rules to enhance ML model performance in clinical note mining.
  • To improve the accuracy of extracting information from clinical notes, particularly when training data is limited.
  • To compare the performance of rule-based systems, ML models, and hybrid approaches.

Main Methods:

  • Collected 501 prostate pathology reports from 6 hospitals, segmented into 2,711 core segments, and annotated with 20 tumor attributes.
  • Developed and evaluated four systems: a rule-based approach, a machine learning (ML) model, a Rule as Feature hybrid model, and a Classifier Confidence hybrid model.
  • Assessed performance using logistic regression (LR), support vector machine (SVM), and eXtreme gradient boosting (XGB) algorithms in a cross-institutional setting.

Main Results:

  • Single-institution training showed LR lagging behind rules (92.2% vs. 95.7% F1 score), while hybrid models achieved competitive results (Classifier Confidence at 96.2%).
  • With multi-institutional data, LR improved (97.2%), but hybrid systems showed greater gains: Rule as Feature (97.7%) and Classifier Confidence (98.3%).
  • Replacing LR with SVM or XGB demonstrated similar performance improvements for the hybrid models.

Conclusions:

  • Developed novel methods to integrate pre-existing handcrafted rules with ML algorithms for clinical note mining.
  • Hybrid systems demonstrated superior performance compared to rule-based or ML models used independently, especially with limited training data.
  • These findings suggest a viable strategy for enhancing clinical text analysis in resource-constrained environments.