A Novel Natural Language Processing Tool Improves Colonoscopy Auditing of Adenoma and Serrated Polyp Detection Rates

Area of Science:

• Gastroenterology
• Medical Informatics
• Health Services Research

Background:

• Manual matching of colonoscopy and histology reports is time-consuming for determining adenoma detection rate (ADR) and serrated polyp detection rate (SDR).
• Automated data extraction methods are needed to improve the efficiency of colonoscopy quality audits.

Purpose of the Study:

• To validate a Natural Language Processing (NLP) code for rapid and efficient extraction of data to calculate ADR and SDR.
• To assess the accuracy and time efficiency of the NLP code in a tertiary health service setting.

Main Methods:

• A NLP code was developed to extract quality indicators from colonoscopy and histology reports.
• Manual review was used to verify the concordance of ADR and SDR extracted by the NLP code.
• The NLP code underwent iterative refinement and validation on separate cohorts.

Main Results:

• The NLP code achieved 99.9% concordance for patient names and 98.9% accuracy for ADR and SDR in the initial training phase.
• After code modification, accuracy improved to 100% in the training cohort and 99.9% in the validation cohort.
• Auditing time in the validation phase using NLP was reduced to less than one hour.

Conclusions:

• An automated NLP code can accurately determine ADR and SDR with nearly 100% accuracy in a tertiary colonoscopy service.
• Wider adoption of NLP offers significant improvements for accurate and time-efficient colonoscopy audits.