Evaluation of a chief complaint pre-processor for biosurveillance

Area of Science:

• Public Health
• Health Informatics
• Epidemiology

Background:

• Emergency Department (ED) chief complaint (CC) data are vital for syndromic surveillance.
• Standardizing varied CC data is challenging due to inconsistent terminology.
• Existing methods struggle with the semantic and lexical diversity of CC data.

Purpose of the Study:

• To revise and evaluate a text processor for pre-processing ED CC data for syndromic surveillance.
• To specifically adapt and assess the processor's effectiveness for acute respiratory illness surveillance.
• To support public health epidemiologists' decision-making through improved data processing.

Main Methods:

• A previously developed text processor was revised.
• The processor was evaluated using 3,699 ED records from a public health surveillance system.
• Accuracy was assessed by comparing syndrome query results against a standard keyword method on raw data.

Main Results:

• The revised text processor demonstrated equal sensitivity, specificity, and predictive values compared to the standard keyword method.
• The processor effectively handled the semantic and lexical variations in CC data.
• Customization for respiratory surveillance yielded comparable performance metrics.

Conclusions:

• The enhanced text processor is a viable tool for standardizing ED chief complaint data for syndromic surveillance.
• This approach supports accurate and efficient tracking of acute respiratory illnesses.
• Improved data processing can enhance public health surveillance and response capabilities.