Redefining Health Care Data Interoperability: Empirical Exploration of Large Language Models in Information Exchange

Area of Science:

• Medical Informatics
• Natural Language Processing
• Artificial Intelligence in Healthcare

Background:

• Healthcare data exchange and interoperability are hindered by nonstandardized and unstructured medical records.
• Large language models (LLMs) offer a potential solution to these information exchange challenges.

Purpose of the Study:

• To evaluate the capability of LLMs in transforming and transferring healthcare data.
• To assess LLM performance in supporting healthcare data interoperability.

Main Methods:

• Three experiments were conducted using data from MIMIC-III and UK Biobank.
• Experiment 1: Assessed LLM accuracy in converting structured lab results to unstructured format.
• Experiment 2: Compared LLM-based diagnostic code conversion (ICD-9-CM to SNOMED-CT) with traditional mapping. Experiment 3: Focused on extracting information from unstructured clinical notes.

Main Results:

• LLM text-based approach demonstrated high accuracy in lab result transformation and improved diagnostic code conversion consistency.
• The LLM achieved an 87.2% positive predictive value for extracting generic drug names from unstructured records.

Conclusions:

• LLMs show significant potential to enhance healthcare data interoperability through accurate and efficient data transformation and exchange.
• LLMs can improve medical data exchange without the need for complex standardization of terms and data structures.