A free-text processing system to capture physical findings: Canonical Phrase Identification System (CAPIS)

Area of Science:

• Medical Informatics
• Clinical Research Data Management
• Natural Language Processing in Healthcare

Background:

• Gathering detailed patient information from free-text medical records is challenging for clinical research.
• Existing methods for extracting physical examination findings are often manual and time-consuming.
• There is a need for automated solutions to streamline data extraction from clinical notes.

Purpose of the Study:

• To develop and evaluate a prototype system for automatically extracting physical examination findings from dictated admission summaries.
• To assess the system's performance in terms of accuracy and efficiency for clinical research applications.
• To demonstrate the utility of concept-based free-text processing for enhancing clinical databases.

Main Methods:

• Developed a prototype computer program utilizing a concept-based free-text processing algorithm.
• The algorithm was designed to identify user-selected target physical examination findings within dictated admission summaries.
• Evaluated the system by comparing its extracted findings against those identified by an independent human investigator.

Main Results:

• The prototype system demonstrated high performance in extracting physical examination findings.
• Achieved a recall (sensitivity) of 92 percent for relevant physical findings.
• Reported a precision (positive predictive value) of 96 percent for the extracted findings.

Conclusions:

• The developed prototype system effectively extracts physical examination findings from free-text medical records.
• The system's high sensitivity and precision suggest its potential to significantly improve the efficiency and accuracy of clinical data enrichment.
• This automated approach offers a valuable tool for advancing clinical research by overcoming barriers in data extraction from unstructured clinical notes.