Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Speaking the Same Language: Leveraging LLMs in Standardizing Clinical Data for AI

Arindam Sett, Somaye Hashemifar, Mrunal Yadav, Yogesh Pandit, Mohsen Hejrati

TL;DR

This paper addresses the problem of heterogeneous clinical data across sources by proposing a large-language-model–driven approach to standardize datasets to HL7 FHIR. It combines Retrieval-Augmented Generation with GPT-3.5 and FAISS-based context retrieval to map data dictionaries to FHIR resources, using a semi-supervised ground-truth process refined by data curators. Evaluation across 14 diverse datasets shows a mean mapping score of 73.54% with high resource alignment (≈94.5%), demonstrating reliable cross-dataset performance and the potential for substantial time and cost savings compared to manual curation. The work suggests that LLMs, when augmented with curated FHIR documentation and careful prompting, can streamline clinical data preparation for AI, improving interoperability and enabling scalable AI deployment in healthcare. It also outlines future work to enhance context, diagnose weak mappings, and compare RAG strategies to further improve robustness and generalizability.

Abstract

The implementation of Artificial Intelligence (AI) in the healthcare industry has garnered considerable attention, attributable to its prospective enhancement of clinical outcomes, expansion of access to superior healthcare, cost reduction, and elevation of patient satisfaction. Nevertheless, the primary hurdle that persists is related to the quality of accessible multi-modal healthcare data in conjunction with the evolution of AI methodologies. This study delves into the adoption of large language models to address specific challenges, specifically, the standardization of healthcare data. We advocate the use of these models to identify and map clinical data schemas to established data standard attributes, such as the Fast Healthcare Interoperability Resources. Our results illustrate that employing large language models significantly diminishes the necessity for manual data curation and elevates the efficacy of the data standardization process. Consequently, the proposed methodology has the propensity to expedite the integration of AI in healthcare, ameliorate the quality of patient care, whilst minimizing the time and financial resources necessary for the preparation of data for AI.

Speaking the Same Language: Leveraging LLMs in Standardizing Clinical Data for AI

TL;DR

This paper addresses the problem of heterogeneous clinical data across sources by proposing a large-language-model–driven approach to standardize datasets to HL7 FHIR. It combines Retrieval-Augmented Generation with GPT-3.5 and FAISS-based context retrieval to map data dictionaries to FHIR resources, using a semi-supervised ground-truth process refined by data curators. Evaluation across 14 diverse datasets shows a mean mapping score of 73.54% with high resource alignment (≈94.5%), demonstrating reliable cross-dataset performance and the potential for substantial time and cost savings compared to manual curation. The work suggests that LLMs, when augmented with curated FHIR documentation and careful prompting, can streamline clinical data preparation for AI, improving interoperability and enabling scalable AI deployment in healthcare. It also outlines future work to enhance context, diagnose weak mappings, and compare RAG strategies to further improve robustness and generalizability.

Abstract

The implementation of Artificial Intelligence (AI) in the healthcare industry has garnered considerable attention, attributable to its prospective enhancement of clinical outcomes, expansion of access to superior healthcare, cost reduction, and elevation of patient satisfaction. Nevertheless, the primary hurdle that persists is related to the quality of accessible multi-modal healthcare data in conjunction with the evolution of AI methodologies. This study delves into the adoption of large language models to address specific challenges, specifically, the standardization of healthcare data. We advocate the use of these models to identify and map clinical data schemas to established data standard attributes, such as the Fast Healthcare Interoperability Resources. Our results illustrate that employing large language models significantly diminishes the necessity for manual data curation and elevates the efficacy of the data standardization process. Consequently, the proposed methodology has the propensity to expedite the integration of AI in healthcare, ameliorate the quality of patient care, whilst minimizing the time and financial resources necessary for the preparation of data for AI.
Paper Structure (17 sections, 2 equations, 2 figures, 4 tables)

This paper contains 17 sections, 2 equations, 2 figures, 4 tables.

Table of Contents

  1. Introduction
  2. A Brief Overview of Data Standardization for Clinical Data
  3. Tasks and Data
  4. Data
  5. Ground-truth Generation and Curation
  6. Methods
  7. System Inspiration
  8. FHIR Standard and Clinical Data Mapping to FHIR
  9. System Overview
  10. Context for RAG
  11. Prompt Engineering
  12. Preliminary Evaluation
  13. Observations
  14. Future Work
  15. Conclusion
  16. ...and 2 more sections

Figures (2)

  • Figure 1: An example of Patient and Media FHIR resources' schemas
  • Figure 2: An overview of our approach.