Novel approach exploring the correlation between presepsin and routine laboratory parameters using explainable...

Area of Science:

• Biomarker Research
• Medical Informatics
• Artificial Intelligence in Medicine

Background:

• Presepsin is a vital sepsis biomarker, but its correlation with routine laboratory and demographic data is not well understood.
• Understanding these relationships is crucial for improving sepsis diagnosis and management.

Purpose of the Study:

• To investigate the relationship between presepsin and routine laboratory parameters using machine learning (ML) and explainable AI (XAI).
• To identify key parameters that enhance the accuracy of sepsis prediction.

Main Methods:

• Employed advanced ML classifiers for data analysis and to highlight interrelationships.
• Utilized XAI to ensure transparency and identify critical parameters for classification.
• Managed missing data using Extreme Gradient Boosting (XGBoost) to maintain data integrity.

Main Results:

• Achieved high predictive accuracy for sepsis patients with an ROC AUC of 0.97 and an accuracy of 0.94.
• XAI successfully identified critical parameters that significantly boosted predictive accuracy.
• XGBoost effectively handled missing data, preserving result accuracy and relevance.

Conclusions:

• The integration of ML and XAI provides novel insights into presepsin's associations with routine clinical data.
• This approach enhances sepsis patient diagnosis and treatment strategies.
• Combining traditional methods with advanced analytics offers significant potential in medical research.