Supporting topic modeling and trends analysis in biomedical literature

Area of Science:

• Biomedical Informatics
• Computational Linguistics
• Data Science

Background:

• Topic modeling, often using Latent Dirichlet Allocation (LDA), extracts key themes from documents.
• Existing LDA implementations typically require programming expertise, limiting accessibility for many professionals.
• Biomedical literature analysis benefits from efficient methods to identify trends and key research areas.

Purpose of the Study:

• To present a user-friendly, web-based application for topic modeling and comparative trends analysis of scientific literature.
• To support biomedical professionals, particularly those without programming skills, in analyzing research literature.
• To facilitate a comprehensive workflow for literature review and scientific discovery.

Main Methods:

• Development of a web-based application integrating topic modeling algorithms.
• Focus on user-friendliness and a complete workflow from data input to trend analysis.
• Usability and efficacy evaluation with 15 biomedical professionals lacking programming expertise.

Main Results:

• The application demonstrated positive acceptance of its functionalities among users.
• An overall usability score of 76/100 was achieved on the System Usability Scale (SUS).
• Users found the system effective for topic modeling and comparative trends analysis.

Conclusions:

• A user-friendly application can significantly increase the adoption of topic modeling by biomedical professionals.
• This approach democratizes advanced literature analysis, opening new avenues for research and review.
• The developed tool enhances the ability to navigate and understand the vast body of scientific literature.