Toward Automated Detection of Documentation Inconsistencies in Electronic Health Records

2026-07-24

AI Summary

This study addresses the critical issue of documentation inconsistencies in electronic health records (EHRs), which can compromise clinical decision-making and patient safety. The authors propose the first hierarchical ontology framework specifically designed for EHR inconsistencies, capturing a spectrum ranging from strict contradictions to ambiguous discrepancies. They introduce a fine-grained annotation schema structured along four axes: category, section, clinical domain, and inconsistency type. Leveraging this framework, they develop a two-stage large language model pipeline: Gemini 2.5 Pro first identifies candidate inconsistencies, followed by context-anchored validation using Gemini 2.5 Flash. Applied to 3,000 discharge summaries from MIMIC-IV-Note, the pipeline detected 3,460 inconsistencies across 69.7% of records, predominantly involving demographics, allergies, diagnoses, and medications, while also exposing systematic model limitations in temporal reasoning and outpatient medication knowledge.

Abstract

Objective: To characterize the kinds of internal documentation inconsistencies a general-domain large language model (LLM) can surface from real-world discharge summaries, and to identify recurring failure modes that limit reliability at scale. Materials and Methods: We applied a two-stage LLM pipeline---open-ended candidate identification (Gemini 2.5 Pro) followed by context-grounded verification (Gemini 2.5 Flash)---to 3,000 randomly sampled MIMIC-IV-Note discharge summaries. A subset of the pipeline output was then reviewed manually by clinical experts. Results: Our pipeline surfaced 3,460 candidate inconsistencies, affecting 69.7% of admissions. Representative examples spanned demographics, allergies, procedures, diagnoses, laboratory, medications, and care-planning domains, with direct implications for clinical reasoning or patient safety. Expert review also revealed recurring failure modes that arise when verification requires temporal reasoning, evolving-diagnosis context, or knowledge of outpatient-prescribing conventions the model does not natively possess. Discussion: Detection is highly context-dependent: many flagged pairs require anchoring each statement to its source section and clinical domain, then assessing whether the conflict reflects a true contradiction or missing context. We propose a graded ontology spanning strict contradiction and ambiguity, with a schema characterizing each flagged case by category, section, domain, and inconsistency axis. Conclusion: This formative study establishes a methodological foundation and conceptual framework to guide subsequent validated, large-scale EHR-inconsistency analysis.

Problem

• documentation inconsistencies
• electronic health records
• discharge summaries
• clinical safety
• natural language processing

Innovation

• large language model
• documentation inconsistency
• electronic health records
• context-grounded verification
• graded ontology