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: Sep 27, 2025

A Knowledge Graph Approach to Elucidate the Role of Organellar Pathways in Disease via Biomedical Reports

A Knowledge Graph Approach to Elucidate the Role of Organellar Pathways in Disease via Biomedical Reports

Published on: October 13, 2023

Effects of data and entity ablation on multitask learning models for biomedical entity recognition.

Nicholas E Rodriguez¹, Mai Nguyen², Bridget T McInnes¹

¹Department of Computer Science, Virginia Commonwealth University, Richmond 23284, USA.

Journal of Biomedical Informatics

|April 12, 2022

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

The NorthStar Trauma Network: An Orthopedic Care Network across Three Health Systems.

NEJM catalyst innovations in care delivery·2026

Same author

Imaging-Guided Live Single-Cell Lipid Profiling of Leader and Follower Cells During Collective Migration of Triple-Negative Breast Cancer Cells.

Advanced science (Weinheim, Baden-Wurttemberg, Germany)·2026

Same author

LRRK2 mutations block NCOA4 trafficking upon iron overload leading to ferroptotic death.

Journal of cell science·2026

Same author

Synthesis of 1'-Homologated-3'-Modified Adenosine Derivatives through Asymmetric Intramolecular S<sub>N</sub>2' Alkylation Using an Evans Chiral Auxiliary.

The Journal of organic chemistry·2026

Same author

Dysfunction of α2δ4 leads to photoreceptor degeneration through disrupted synaptic mitochondria and calcium crosstalk.

Cell death & disease·2026

Same author

Psychological foundations of ambiguity in the hybrid workplace: The role of managerial risk-taking and AI-induced job insecurity.

Acta psychologica·2026

Multi-task learning for named entity recognition (NER) models can train new tasks with less data. While not always improving performance, it enables efficient model development for biomedical applications.

Area of Science:

Natural Language Processing (NLP)
Machine Learning
Bioinformatics

Background:

Training domain-specific Named Entity Recognition (NER) models demands extensive, costly, hand-curated datasets.
Deploying numerous NLP models presents significant storage and memory challenges.

Purpose of the Study:

To investigate the efficacy of multi-task learning (MTL) in reducing the training data requirements for new domain-specific NER models.
To assess the impact of transfer learning on task performance across diverse biomedical NER datasets.

Main Methods:

Utilized multi-task learning to train models on multiple biomedical NER datasets simultaneously.
Evaluated performance across 22 distinct biomedical NER datasets.
Employed ablation studies to quantify the contribution of transfer learning.

Keywords:

Biomedical text processing Deep learning Machine learning Named entity recognition Natural language processing

Related Experiment Videos

Last Updated: Sep 27, 2025

A Knowledge Graph Approach to Elucidate the Role of Organellar Pathways in Disease via Biomedical Reports

A Knowledge Graph Approach to Elucidate the Role of Organellar Pathways in Disease via Biomedical Reports

Published on: October 13, 2023

Main Results:

Multi-task learning models generally performed comparably to single-task models, without significant performance gains.
Demonstrated that initializing new, unseen tasks with weights from a multi-task model can enable training with less data and improve performance in specific scenarios.

Conclusions:

Multi-task learning offers a viable approach for developing new NER models with reduced data requirements, particularly when leveraging pre-trained weights.
While not universally superior, MTL provides a valuable strategy for efficient model development in data-scarce biomedical domains.