Fourier transform infrared spectroscopy for Streptococcus pneumoniae capsular serotype classification in pediatric...

Area of Science:

• Microbiology
• Spectroscopy
• Machine Learning

Background:

• Invasive pneumococcal disease (IPD) poses a significant global health burden, especially in children.
• Current serotyping methods for Streptococcus pneumoniae are time-consuming and complex.
• Rapid and accurate serotyping is crucial for effective IPD management and surveillance.

Purpose of the Study:

• To develop and validate a Fourier-transform infrared (FT-IR) spectroscopy-based classifier for Streptococcus pneumoniae serotyping.
• To establish FT-IR spectroscopy as a rapid, cost-effective alternative to traditional serotyping techniques.
• To leverage artificial neural network (ANN) machine learning for enhanced classification accuracy.

Main Methods:

• Collected 76 clinical isolates of Streptococcus pneumoniae from pediatric IPD patients.
• Performed traditional Quellung reaction serotyping for all isolates.
• Acquired 843 FT-IR spectra from the isolates.
• Developed an artificial neural network (ANN) model for serotype classification using FT-IR spectra.
• Validated the ANN model internally using separate training and testing datasets.

Main Results:

• The FT-IR spectroscopy and ANN model achieved a predictive accuracy of 98% for several key serotypes.
• The classifier demonstrated high accuracy for serotypes including 19A, 6, 3, 14, 18C, 22F, 23A, 23B, 33F, 35B, and 9N.
• Certain serotype groups (10A/16F, 15ABC, 7CF) required grouping due to differentiation limitations.

Conclusions:

• FT-IR spectroscopy is a promising, rapid, and low-cost method for phenotypic classification of Streptococcus pneumoniae serotypes.
• This methodology holds significant implications for clinical practice, patient management, and infection trend monitoring.
• The technique supports the development of new vaccines and improves epidemiological surveillance of IPD.