Enhancing Patient Safety Event Reporting by K-nearest Neighbor Classifier

Area of Science:

• Health Informatics
• Computer Science
• Patient Safety

Background:

• Patient safety event reporting systems often suffer from low utility due to poor data quality.
• Human factors in data entry significantly impact data quality, necessitating research into improved methods.
• The choice between structured and unstructured data entry presents trade-offs in accuracy, completeness, and timeliness, alongside text mining challenges.

Purpose of the Study:

• To address the critical need for improved data quality in patient safety event reporting.
• To investigate the application of text classification for enhancing data categorization and utility.
• To evaluate the k-nearest neighbor (KNN) algorithm for predicting subject categories in a reporting system.

Main Methods:

• A text classification approach using the k-nearest neighbor (KNN) algorithm was proposed.
• The KNN classifier was applied to predict subject categories within a patient safety event reporting system.
• The study focused on improving data entry and leveraging text mining techniques.

Main Results:

• The feasibility of using the KNN classifier for text classification in this context was demonstrated.
• The application of KNN showed potential for improving data quality in patient safety event reporting.
• The method indicated advantages for enhancing clinical decision support.

Conclusions:

• The k-nearest neighbor (KNN) text classification method is a viable approach for improving patient safety event reporting.
• Implementing such text classification can significantly enhance data quality and support better clinical decision-making.
• Addressing data quality through advanced methods like KNN is crucial for maximizing the utility of safety reporting systems.