Exploring Hidden In-Hospital Fall Clusters from Incident Reports Using Text Analytics.

Area of Science:

• Healthcare Quality Improvement
• Data Mining
• Patient Safety

Background:

• Retrospective analysis of fall incident reports is crucial for uncovering hidden information and identifying risk factors.
• Improving healthcare quality necessitates understanding the nuances of patient falls.

Purpose of the Study:

• To explore potential fall incident clusters using text mining techniques.
• To categorize fall reports to identify distinct patterns and risk factors.

Main Methods:

• Utilized word embeddings and hierarchical clustering on fall incident reports from 7 Hong Kong hospitals.
• Catalogued reports into 5 distinct clusters based on analyzed data.

Main Results:

• Identified 5 clusters with significant differences in fall severity, patient gender, and reporting departments.
• Keywords associated with each cluster provided further insights into fall characteristics.

Conclusions:

• Demonstrated the feasibility of applying text clustering methods to real-world fall incident reports.
• This approach offers a novel way to mine valuable information from incident reports for enhanced patient safety.