Literature mining method RaJoLink for uncovering relations between biomedical concepts

Area of Science:

• Biomedical informatics
• Computational biology
• Knowledge discovery

Background:

• Biomedical research generates vast amounts of literature, making comprehensive knowledge discovery challenging.
• Identifying novel relationships between disparate biomedical concepts is crucial for advancing scientific understanding.
• Existing literature mining tools may not effectively bridge disconnected sets of articles.

Purpose of the Study:

• To introduce RaJoLink, a semi-automated literature mining method for uncovering hidden relations between biomedical concepts.
• To implement Swanson's ABC model approach for hypothesis generation in a novel way.
• To facilitate knowledge discovery for biomedical experts by connecting seemingly unrelated research areas.

Main Methods:

• RaJoLink semi-automatically suggests candidate 'agents' (a) linked to a phenomenon (c) using rare terms found in literature on 'c'.
• Identifies a common term linking literature on rare terms associated with 'c' as a candidate for 'a'.
• Searches for linking terms ('b') between literature on 'a' and 'c' to provide supporting evidence for discovered connections.

Main Results:

• The RaJoLink method was applied to the biomedical literature on autism, utilizing MEDLINE as the data source.
• Expert evaluation confirmed the potential of the discovered relations to enhance the understanding of autism.
• The method successfully identified potential connections between concepts within disconnected sets of articles.

Conclusions:

• RaJoLink offers a novel approach to literature mining, supporting biomedical knowledge discovery.
• The method's ability to identify unusual observations via rare terms aids in hypothesis generation.
• Discovered relations show promise for advancing the understanding of complex diseases like autism.