Efficient Identification of Existing Definitions for Improving Interoperability of Electronic Case Report Forms

Area of Science:

• Clinical informatics
• Data management
• Natural language processing

Background:

• Efficient retrieval of clinical concept definitions and terminology codes is crucial for data standardization in electronic case report forms (eCRF).
• Current methods for searching within eCRF databases may be suboptimal.
• Comparing different search strategies is needed to improve efficiency.

Purpose of the Study:

• To investigate efficient methods for searching clinical concept definitions and terminology codes.
• To compare the effectiveness of clinical concept definition databases versus eCRF databases for search.
• To evaluate a large language model (LLM)-assisted search approach.

Main Methods:

• Comparative analysis of search strategies.
• Utilizing clinical concept definition databases.
• Utilizing eCRF databases.
• Implementing and assessing a large language model (LLM)-assisted search.

Main Results:

• The large language model (LLM) approach demonstrated enhanced efficiency in searching for clinical concept definitions and terminology codes.
• Direct searching within eCRF databases presented challenges compared to specialized definition databases.
• LLM-assisted search facilitated quicker identification of relevant concepts.

Conclusions:

• A large language model (LLM) offers a promising and efficient solution for searching clinical concept definitions and terminology codes.
• LLM-assisted search can improve data standardization and retrieval within clinical research.
• Future research should explore further integration of LLMs in clinical data management.