Performance of Combined Support Vector Machine and Principal Component Analysis in recognizing infant cry with...

Area of Science:

• Biomedical Engineering
• Signal Processing
• Machine Learning in Healthcare

Background:

• Infant cry analysis is crucial for diagnosing conditions like asphyxia.
• Distinguishing pathological cries from healthy ones requires sophisticated methods.
• Traditional methods may lack the accuracy needed for early detection.

Purpose of the Study:

• To develop and evaluate a machine learning model for classifying infant cries associated with asphyxia.
• To investigate the effectiveness of combining Principal Component Analysis (PCA) with Support Vector Machine (SVM) for cry classification.
• To assess the performance of different SVM kernels in this diagnostic task.

Main Methods:

• Utilized a combined approach of Support Vector Machine (SVM) and Principal Component Analysis (PCA).
• PCA was employed for feature selection and dimensionality reduction of cry audio signals.
• An SVM classifier was trained on PCA-selected features to differentiate between healthy and asphyxiated infant cries.
• Evaluated SVM performance using both linear and Radial Basis Function (RBF) kernels.

Main Results:

• The combined SVM-PCA model demonstrated high performance in classifying infant cries.
• The SVM classifier with an RBF kernel achieved a classification accuracy of 95.86%.
• PCA effectively reduced the input data dimensionality for the SVM, enhancing efficiency.

Conclusions:

• The SVM-PCA method is a highly effective tool for the objective classification of infant cries related to asphyxia.
• The RBF kernel within the SVM framework provides superior performance for this specific application.
• This approach holds potential for improving early diagnosis and intervention for infants with asphyxia.