Ontology-enhanced automatic chief complaint classification for syndromic surveillance

Area of Science:

• Public Health
• Medical Informatics
• Natural Language Processing

Background:

• Emergency department chief complaints (CCs) are vital for public health surveillance.
• Current CC classification methods face challenges due to vocabulary variations and lack of standardized encoding.
• Effective classification is crucial for automated analysis and timely public health insights.

Purpose of the Study:

• To develop and evaluate an ontology-enhanced automatic classification approach for emergency department chief complaints (CCs).
• To address the problem of vocabulary variation in CCs.
• To create a flexible classification system capable of handling multiple syndromic category sets.

Main Methods:

• Developed an automatic CC classification approach leveraging semantic relations within a medical ontology.
• Compared the proposed ontology-enhanced method against two popular CC classification techniques.
• Utilized a real-world dataset of emergency department chief complaints for experimental validation.

Main Results:

• The ontology-enhanced approach demonstrated significantly superior performance compared to benchmark methods.
• Improvements were observed in key performance metrics including sensitivity, F measure, and F2 measure.
• The method effectively handles the inherent vocabulary variations in free-text chief complaints.

Conclusions:

• Ontology-enhanced automatic classification offers a robust solution for syndromic surveillance using emergency department chief complaints.
• This approach improves the accuracy and reliability of automated public health data analysis.
• The developed method provides a flexible and effective tool for classifying CCs across diverse syndromic surveillance needs.