A Machine Learning-Based Risk Score for Prediction of Infective Endocarditis Among Patients With Staphylococcus...

Area of Science:

• Infectious Diseases
• Cardiology
• Data Science

Background:

• Stratifying Staphylococcus aureus infective endocarditis (SA-IE) risk in patients with S. aureus bacteremia (SAB) is crucial for timely clinical management.
• Existing risk assessment methods often rely on subjective clinical judgment, necessitating objective tools for early risk stratification.

Purpose of the Study:

• To develop and validate a novel, objective risk score for predicting SA-IE in SAB patients.
• The risk score aims for early application at the time of blood culture positivity, independent of clinical judgment.

Main Methods:

• A retrospective big data analysis of hospitalized SAB patients (2009-2019) was performed.
• A random forest model identified key predictive variables for SA-IE.
• Data were split into derivation and validation cohorts, with AUCROCs calculated.

Main Results:

• The study included 15,741 SAB patients, with 4.18% diagnosed with SA-IE.
• The developed risk score achieved an AUCROC of 0.74, demonstrating good predictive performance.
• Key predictors identified were age, prior infective endocarditis, valvular heart disease, and community onset.

Conclusions:

• A novel, objective risk score for SA-IE in SAB patients was successfully developed.
• The risk score's performance is comparable to existing methods and can be applied early in patient management.
• This tool facilitates early risk stratification, aiding clinical decision-making upon SAB diagnosis.