Improving Prediction Accuracy of "Central Line-Associated Blood Stream Infections" Using Data Mining Models.

Area of Science:

• Medical Informatics
• Data Mining in Healthcare
• Clinical Surveillance

Background:

• Nosocomial infections pose significant challenges to patient safety and healthcare costs.
• Developing effective clinical surveillance programs for infection prediction is complex due to high-dimensional and heterogeneous medical data.
• Accurate prediction of infections like central line-associated bloodstream infections (CLABSIs) is vital for timely preventive actions.

Purpose of the Study:

• To evaluate data mining methods for predicting central line-associated bloodstream infections (CLABSIs).
• To identify the most effective data mining technique for enhancing clinical surveillance programs.
• To improve the accuracy of CLABSI detection and support infection control initiatives.

Main Methods:

• Implementation of six data mining methods using the Cross Industry Standard Process for Data Mining (CRISP-DM).
• Utilization of healthcare-associated infection datasets from the US National Healthcare Safety Network.
• Integration of consumer survey data from the Hospital Consumer Assessment of Healthcare Providers and Systems.

Main Results:

• The AdaBoost method demonstrated the highest accuracy, reaching up to 89.7% in predicting CLABSIs.
• Successful prediction of CLABSIs using the selected data mining approaches.
• Validation of the potential for data mining in improving infection detection rates.

Conclusions:

• AdaBoost is a highly effective method for predicting CLABSIs, significantly improving clinical surveillance capabilities.
• Accurate CLABSI prediction can lead to reduced patient hospital stay costs and enhanced patient safety.
• The findings support the integration of advanced data mining techniques into healthcare infection control strategies.