Ability of Procalcitonin and C-Reactive Protein for Discriminating between Bacterial and Enteroviral Meningitis in Children Using Decision Tree

Area of Science:

• Infectious Diseases
• Pediatrics
• Medical Diagnostics

Background:

• Bacterial meningitis (BM) poses a significant public health challenge, particularly in developing regions like Central Asia, due to high mortality and neurological sequelae.
• Delayed treatment initiation in acute bacterial meningitis correlates with increased mortality rates.
• Current diagnostic methods, such as cerebrospinal fluid culture, are slow and have limited sensitivity, hindering prompt clinical management.

Purpose of the Study:

• To develop and validate a rapid diagnostic method for distinguishing bacterial meningitis from enteroviral meningitis (EVM) in pediatric patients.
• To leverage machine learning to identify key biomarkers and establish decision rules for accurate meningitis differentiation.
• To improve the timeliness and efficiency of meningitis diagnostics in resource-limited settings.

Main Methods:

• A machine learning approach utilizing fast-and-frugal trees (FFTree) decision trees was employed.
• The study focused on analyzing procalcitonin and C-reactive protein (CRP) levels in cerebrospinal fluid and blood.
• Cut-off values for these biomarkers were determined to differentiate between bacterial and enteroviral meningitis.

Main Results:

• The FFTree model successfully identified procalcitonin and CRP cut-off values for distinguishing bacterial from enteroviral meningitis in children.
• The proposed method achieved exceptional diagnostic performance: 100% sensitivity, 96% specificity, and 98% accuracy.
• These results indicate a highly reliable differentiation between the two types of meningitis.

Conclusions:

• The developed FFTree-based method provides a fast, accurate, and potentially cost-effective tool for differentiating bacterial from enteroviral meningitis.
• This approach can significantly aid clinicians in making timely treatment decisions, potentially reducing mortality and morbidity associated with bacterial meningitis.
• Optimizing meningitis diagnostics through rapid, reliable methods is crucial for improving patient care, especially in regions with high disease burden.