Jackalope Plus tool for post-coordination, ontology development, and precise mapping in observational health studies

Area of Science:

• Health Informatics
• Artificial Intelligence in Healthcare

Background:

• Mapping complex health data to the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) presents significant challenges in maintaining clinical accuracy.
• Existing tools often struggle with the nuance of intricate medical terminologies, leading to data standardization inefficiencies.

Purpose of the Study:

• To introduce Jackalope Plus, a novel tool designed to enhance the precision and efficiency of mapping complex health data to the OMOP CDM.
• To leverage SNOMED CT post-coordination and a GPT-4o mini large language model (LLM) for improved clinical concept standardization.

Main Methods:

• A two-step approach combining semantic search with LLM-driven standardization was employed.
• The tool utilizes SNOMED CT post-coordination and a GPT-4o mini LLM to process and standardize intricate medical concepts.

Main Results:

• Jackalope Plus achieved over 77.5% accuracy in mapping complex terminologies on benchmark and custom datasets.
• The tool outperformed Usagi (52.5% accuracy) and matched manual mapping accuracy while reducing processing time by up to 50%.

Conclusions:

• Jackalope Plus offers a versatile and highly accurate solution for standardizing diverse healthcare data within the OMOP CDM.
• Future development will focus on user feedback and resolving concept ambiguities, with a beta version available for research.