PubMedMiner automatically retrieves a set of very recent PubMed articles for a given query word and facilitates further analysis that includes: named entity recognition and co-occurrence analysis. The application also builds a co-occurrence network which when visualized gives an idea about the relation between the query word and different biomedical concepts identified in the texts using different ontologies. Such analysis is especially helpful in relating an entity (for example a drug) of recent interest to its related biomedical concepts.
The following screenshots of the application give an idea about how the tool looks and what it can do.
Fig 1: Basic screenshot displaying 10 PubMed Abstracts for "diclofenac" with the chemical names in the text tagged (using OSCAR4 dictionary) in Yellow color.
Fig 2: A screenshot of the co-occurrence network created using the application for those chemical names which co-occur along with the query ("diclofenac").
Fig 3: A screenshot of the co-occurrence matrix or correlation matrix created using the application for the same as Fig 2. Darkly colored cells indicate more frequent co-occurrence of the corresponding terms.
- Note: The application currently retrieves only a maximum of 299 Abstracts per run due to the limitations in fetching higher number of Abstracts from PubMed with a single http request. The application will be developed further.