Multi-Component Pipeline LLMs for Interoperable Healthcare Data: A Scoping Review from Clinical Summarization to Multimodal Integration
Year: 2026
Type: book-chapter
Abstract: Healthcare data remain fragmented across structured electronic health record fields, unstructured narratives, imaging metadata, and continuous Internet of Medical Things (IoMT) streams, limiting continuity of care and interoperability. Large language models (LLMs) can support clinical documentation and workflow tasks, but single-model deployments are constrained by hallucinations, omissions, limited provenance, and weak alignment with standards-based exchange. Multi-Component Pipeline LLMs (MCP LLMs)—modular, tool-augmented, and multi-agent architectures that combine retrieval, structured extraction, validation, and governance controls—have emerged as a candidate approach for safer end-to-end patient data management. This scoping review synthesizes recent evidence on MCP LLM architectures for interoperable healthcare data management across four areas: clinical summarization and documentation support; privacy-preserving interaction with EHRs and workflow automation; interoperability enablement through HL7 Fast Healthcare Interoperability Resources (FHIR) mapping, profile/version mediation, and terminology normalization; and multimodal integration linking text with imaging and IoMT/wearable signals. Across 49 included studies, MCP LLM designs emphasized modular controllability via retrieval-augmented generation, tool use, and orchestration, with applications spanning discharge and encounter summary drafting, dialog summarization, and EHR-facing assistance. Interoperability-focused work reported transforming narrative documentation into FHIR-aligned structured representations and terminology-aware normalization, while multimodal studies linked text with imaging and wearable streams for context-aware decision support. Persistent limitations were noted in factual reliability for high-stakes summaries, inconsistent evaluation practices, and the need for robust governance for regulated deployment. Future work should standardize safety-oriented evaluation and interoperability validation alongside task performance.
Source: IntechOpen eBooks
Language: English
Subfield: Health Informatics
Field: Medicine
Domain: Health Sciences
Open Access status: hybrid
OpenAlex ID: W7204252864