Machine Learning for Clinical Decision Support of Acute Streptococcal Pharyngitis: A Pilot Study

Area of Science:

• Pediatric Infectious Diseases
• Medical Informatics
• Machine Learning in Healthcare

Background:

• Group A Streptococcus (GAS) is a common cause of bacterial pharyngitis in children.
• Differentiating GAS from viral pharyngitis is challenging, leading to potential overuse of antibiotics.
• Unnecessary antibiotic use can result in adverse effects and increased antibiotic resistance.

Purpose of the Study:

• To evaluate the effectiveness of a machine learning algorithm in diagnosing bacterial pharyngitis in pediatric patients.
• To assess the algorithm's impact on clinical decision-making for GAS infections.

Main Methods:

• A cohort of 54 children (aged 2-17) with sore throat and fever were evaluated.
• Standard diagnostic tests including RADT and throat culture were performed.
• A linear support vector machine algorithm was applied for classification of GAS-P and viral pharyngitis.

Main Results:

• The machine learning algorithm achieved a positive predictive value of 80.6% for identifying GAS-P infections.
• The false discovery rate for GAS-P infection was 19.4%.

Conclusions:

• Machine learning demonstrates high predictive value in detecting streptococcal pharyngitis.
• This AI strategy can serve as a valuable decision support tool for clinicians managing GAS-P in children.