A computational framework for converting textual clinical diagnostic criteria into the quality data model.

Area of Science:

• Clinical informatics
• Computational linguistics
• Health data standardization

Background:

• Standardizing clinical diagnostic criteria is crucial but challenging in clinical informatics.
• The Quality Data Model (QDM) offers a promising approach for standardizing clinical diagnostic criteria.

Purpose of the Study:

• To develop and evaluate automated methods for converting textual clinical diagnostic criteria into a structured format using QDM.
• To enhance the computability and standardization of clinical diagnostic criteria.

Main Methods:

• Utilized a clinical Natural Language Processing (NLP) tool (cTAKES) for sentence detection and event annotation.
• Developed a rule-based approach for QDM datatype assignment and a Conditional Random Fields (CRFs) machine learning algorithm for attribute annotation.
• Created a manually annotated corpus as a gold standard for performance evaluation using precision, recall, and f-measure.

Main Results:

• Successfully harvested 267 criteria (Symptom, Laboratory Test) from 63 textual criteria.
• Achieved high performance metrics: rule-based approach (0.85 f-measure) and CRFs (0.91 f-measure).
• Developed a web-based tool for automatic translation of Laboratory Test criteria into QDM XML format, demonstrating effectiveness.

Conclusions:

• The developed NLP-based computational framework is a feasible and effective solution for clinical diagnostic criteria representation and computerization.
• Automated conversion of textual criteria to QDM format facilitates data standardization and clinical informatics research.