Synonym, topic model and predicate-based query expansion for retrieving clinical documents.

Area of Science:

• Medical Informatics
• Information Retrieval
• Natural Language Processing

Background:

• Retrieving relevant documents from large clinical data warehouses is challenging.
• Effective information retrieval is crucial for clinical decision-making and research.

Purpose of the Study:

• To develop and evaluate three distinct query expansion methods for clinical document retrieval.
• To compare the performance of synonym expansion, topic modeling, and predicate database methods against a baseline.

Main Methods:

• Implemented synonym expansion using selected vocabularies.
• Trained a topic model on clinical documents to identify related terms.
• Utilized a predicate database from Medline abstracts for term identification.

Main Results:

• All three query expansion methods improved average recall and average F-measure compared to the baseline.
• Average precision and precision at 10 decreased across all expansion methods.
• The topic model-based query expansion demonstrated superior performance in recall and F-measure.

Conclusions:

• Query expansion techniques can enhance the retrieval of clinical documents.
• Topic modeling offers a promising approach for improving recall and F-measure in clinical information retrieval.
• Further research may be needed to balance precision and recall trade-offs.