Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Semantic NLP Pipelines for Interoperable Patient Digital Twins from Unstructured EHRs

Rafael Brens, Yuqiao Meng, Luoxi Tang, Zhaohan Xi

TL;DR

The paper addresses the challenge of converting unstructured EHR content into interoperable patient digital twins using a semantic NLP pipeline aligned to the FHIR standard. It combines transformer-based NER, ontology-grounded normalization, and relation extraction to build minimal FHIR-compliant digital twins. Evaluation on the MIMIC-IV Demo with reference mappings shows significant gains in entity and relation extraction, semantic completeness, and interoperability. The work demonstrates the feasibility of end-to-end automated mapping from free-text to standards-based digital twins and outlines directions for extending longitudinal, multi-modal, and cross-institution generalization.

Abstract

Digital twins -- virtual replicas of physical entities -- are gaining traction in healthcare for personalized monitoring, predictive modeling, and clinical decision support. However, generating interoperable patient digital twins from unstructured electronic health records (EHRs) remains challenging due to variability in clinical documentation and lack of standardized mappings. This paper presents a semantic NLP-driven pipeline that transforms free-text EHR notes into FHIR-compliant digital twin representations. The pipeline leverages named entity recognition (NER) to extract clinical concepts, concept normalization to map entities to SNOMED-CT or ICD-10, and relation extraction to capture structured associations between conditions, medications, and observations. Evaluation on MIMIC-IV Clinical Database Demo with validation against MIMIC-IV-on-FHIR reference mappings demonstrates high F1-scores for entity and relation extraction, with improved schema completeness and interoperability compared to baseline methods.

Semantic NLP Pipelines for Interoperable Patient Digital Twins from Unstructured EHRs

TL;DR

The paper addresses the challenge of converting unstructured EHR content into interoperable patient digital twins using a semantic NLP pipeline aligned to the FHIR standard. It combines transformer-based NER, ontology-grounded normalization, and relation extraction to build minimal FHIR-compliant digital twins. Evaluation on the MIMIC-IV Demo with reference mappings shows significant gains in entity and relation extraction, semantic completeness, and interoperability. The work demonstrates the feasibility of end-to-end automated mapping from free-text to standards-based digital twins and outlines directions for extending longitudinal, multi-modal, and cross-institution generalization.

Abstract

Digital twins -- virtual replicas of physical entities -- are gaining traction in healthcare for personalized monitoring, predictive modeling, and clinical decision support. However, generating interoperable patient digital twins from unstructured electronic health records (EHRs) remains challenging due to variability in clinical documentation and lack of standardized mappings. This paper presents a semantic NLP-driven pipeline that transforms free-text EHR notes into FHIR-compliant digital twin representations. The pipeline leverages named entity recognition (NER) to extract clinical concepts, concept normalization to map entities to SNOMED-CT or ICD-10, and relation extraction to capture structured associations between conditions, medications, and observations. Evaluation on MIMIC-IV Clinical Database Demo with validation against MIMIC-IV-on-FHIR reference mappings demonstrates high F1-scores for entity and relation extraction, with improved schema completeness and interoperability compared to baseline methods.
Paper Structure (23 sections, 1 figure, 3 tables)

This paper contains 23 sections, 1 figure, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Digital Twins in Healthcare.
  4. Clinical NLP.
  5. FHIR and Interoperability.
  6. Methodology
  7. FHIR Digital Twin Profile.
  8. Semantic NLP for Extraction and Normalization.
  9. Relation-Aware Assembly and Validation.
  10. Experiments
  11. Setup and Main Results
  12. Datasets.
  13. Baselines.
  14. Metrics.
  15. Ablation Study
  16. ...and 8 more sections

Figures (1)

  • Figure 1: Overview of the semantic NLP pipeline with illustrative example. Given clinical text (left), the pipeline extracts entities via NER, normalizes them to standard ontologies (SNOMED-CT, RxNorm), identifies relations between entities, and assembles validated FHIR resources into a patient digital twin (right).