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 Experiment Videos

DICE: A Drug Indication Classification and Encyclopedia for AI-Based Indication Extraction.

Arjun Bhatt1,2,3, Ruth Roberts4,5, Xi Chen1

  • 1Division of Bioinformatics & Biostatistics, National Center for Toxicological Research, Food and Drug Administration, Jefferson, AR, United States.

Frontiers in Artificial Intelligence
|August 19, 2021
PubMed
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

Therapists' Experiences of Delivering Therapy to Children Seeking Refuge and Asylum: A Qualitative Study.

Clinical child psychology and psychiatry·2026
Same author

Demographic and Comorbidity Associations With Disease Severity and Bony Dehiscence in AFRS.

Laryngoscope investigative otolaryngology·2026
Same author

Safety considerations for cereblon-recruiting targeted protein degraders.

Nature reviews. Drug discovery·2026
Same author

Association of the Change in USMLE Step 1 Scoring With Research Output Among Applicants to Anesthesiology Residency Programs: A Retrospective Observational Study.

The journal of education in perioperative medicine : JEPM·2026
Same author

Cohort Profile: Our Future-the longitudinal study of young people in England born in 1998-99.

International journal of epidemiology·2026
Same author

Dimensional Associations Between Conduct Problems and Brain Structure Across 18 International Cohorts in ENIGMA.

Journal of the American Academy of Child and Adolescent Psychiatry·2026
This summary is machine-generated.

A new dataset, DICE, aids AI in extracting drug indications from labeling. Transformer models achieved high accuracy, outperforming BiLSTM, and showed promise for drug repositioning and real-world evidence generation.

Area of Science:

  • Computational biology
  • Natural Language Processing
  • Pharmacology

Background:

  • Drug labeling is crucial for clinical decisions and regulatory oversight.
  • Extracting drug indications from text can support drug repositioning and real-world evidence.
  • AI models require robust datasets for accurate information extraction.

Purpose of the Study:

  • To develop a dataset for training AI models to extract drug indications.
  • To evaluate the performance of various AI classifiers for indication extraction.
  • To assess the utility of the developed dataset in real-world applications.

Main Methods:

  • Manual curation and development of the Drug Indication Classification and Encyclopedia (DICE) dataset from FDA drug labeling.
  • Creation of a DICE scheme with 7,231 sentences across five classes.
Keywords:
artificial intelligencedeep learningdrug indicationnatural language processingtransformers

Related Experiment Videos

  • Development and evaluation of nine AI-based classifiers, including transformer and BiLSTM models.

    • Main Results:

    • Transformer-based models achieved an average MCC of 0.887, outperforming BiLSTM models (0.862).
    • The best classifiers achieved a high enrichment rate (>0.930) when extracting indications from DrugBank.
    • Domain-specific training enhanced model explainability and generalization without performance loss.

    Conclusions:

    • The DICE dataset is a valuable resource for developing and evaluating AI-powered NLP models for drug indication extraction.
    • AI models, particularly transformers, show significant potential for automating the extraction of critical drug information.
    • This work facilitates drug repositioning and the secondary use of medicines through improved data extraction.