Tracing Clinical Pathways from Unstructured Medical Documents: A Case Study on a Standard Operating Procedure for Chest Pain

Research output: Chapter in Book/Report/Conference proceedingsConference contributionpeer-review

Abstract

Healthcare systems aim to deliver consistent, high-quality
care through standardized clinical guidelines. However, real-world ad-
herence remains difficult due to dynamic clinical environments and in-
consistent documentation. We present a pipeline that transforms free-
text medical records into HL7 FHIR AuditEvent resources, enabling
process mining (PM) and conformance checking for real-world care path-
ways. Using semantic extraction and rule-based mapping, we reconstruct
chest-pain treatment workflows from unstructured documents and con-
vert them into XES logs for PM at three abstraction levels: metadata-
based, content-informed, and value-driven decision tracking. Applied to
90 documents from 10 patients, our approach recovers complete clini-
cal trajectories, fills documentation gaps, and captures decision-making
steps aligned with SOP thresholds. This demonstrates the feasibility of
utilizing unstructured data for quantitative guideline compliance analysis
and provides a foundation for broader studies on time-sensitive clinical
processes.
Original languageEnglish
Title of host publicationEUROCAST 2026 ComputerAided Systems Theory EXTENDED ABSTRACTS
Publication statusPublished - Feb 2026

Fingerprint

Dive into the research topics of 'Tracing Clinical Pathways from Unstructured Medical Documents: A Case Study on a Standard Operating Procedure for Chest Pain'. Together they form a unique fingerprint.

Cite this