Stratification of Group A Streptococcal Pharyngitis Children Using Unsupervised Learning

Area of Science:

• Pediatric infectious diseases
• Clinical informatics
• Machine learning in healthcare

Background:

• Group A Streptococcus (GAS) is a common cause of bacterial pharyngitis.
• Current diagnostic criteria for GAS pharyngitis rely on pediatrician judgment, lacking clear definition.
• Rapid antigen detection testing (RADT) is recommended for selective use in GAS pharyngitis diagnosis.

Purpose of the Study:

• To apply unsupervised learning to classify GAS pharyngitis cases based on clinical symptoms.
• To identify specific clinical indicators for further investigation and treatment of GAS pharyngitis.
• To improve the diagnostic accuracy of GAS pharyngitis, especially in young children.

Main Methods:

• Analysis of categorical data from 305 RADT-positive patients aged 3-15 years.
• Utilized K-modes clustering to group patients based on clinical manifestations.
• Statistical examination of variable relationships with clusters and differences between clusters.

Main Results:

• K-modes clustering identified two distinct patient groups.
• Cluster 1: Older children with lymph node tenderness.
• Cluster 2: Younger children presenting with cough and rhinorrhea.

Conclusions:

• Distinguishing GAS pharyngitis from common respiratory infections based solely on symptoms is difficult, particularly in young children.
• Further research is needed to identify reliable indicators for suspecting streptococcal infections in pediatric patients.
• Clinical presentation varies significantly with age in GAS pharyngitis, necessitating tailored diagnostic approaches.