An Efficient, Clinically-Natural Electronic Medical Record System that Produces Computable Data.

Area of Science:

• Health Informatics
• Clinical Documentation Improvement
• Natural Language Processing in Healthcare

Background:

• Current electronic medical record (EMR) systems primarily generate text-based data, focusing on financial and regulatory aspects.
• Existing EMR systems lack efficient methods for converting unstructured clinical notes into computable data.
• There is a need for EMR systems that support clinical decision-making and generate actionable data.

Purpose of the Study:

• To develop and evaluate a novel approach for enhancing the computability of clinical documentation within EMR systems.
• To integrate clinical decision support (CDS) with activity-based design principles for improved data capture.
• To demonstrate the efficiency of natural language processing (NLP) in converting structured clinical text into computable data.

Main Methods:

• Combined activity-based design with Care Process Models for integrating CDS into clinical workflows.
• Utilized NLP techniques to process textual clinical documentation.
• Conducted a trial using radiology reports for pneumonia rule-out examinations.

Main Results:

• Over 98% of documentation from a trial involving radiology reports was successfully captured as computable data.
• Subjective evidence suggests similar high data capture rates across various clinical applications (physician, nursing, physical therapy).
• The system enables direct clinician control over clinical content, bypassing IT intermediaries.

Conclusions:

• The developed system efficiently converts clinical documentation into computable data using NLP.
• This approach supports secondary functions like activity-based costing and clinical coding (CPT, ICD-10, SNOMED).
• Widely available computable data holds significant potential for improving healthcare management and patient outcomes.