Development of a Classification Model for Pediatric Cardiac Abnormality Detection Through the Integration of Child Physiological Data and Phonocardiogram Features

EasyChair Preprint 15430

Abstract

Heart murmurs play a truly crucial role in the assessment of cardiac conditions. This particular study undertakes the classification of heart murmurs in patients by making use of Support Vector Machines (SVM) derived from Phonocardiogram (PCG) signals. It employs the CirCor DigiScope dataset, with a specific focus on "Child" subjects. Owing to the limited presence of diastolic murmurs, this study confines its assessment solely to systolic murmurs. The dataset in question encompasses a total of 664 subjects. The recordings within this dataset are carefully segmented into systolic segments. Mel spectrograms and their derivatives are utilized for the purpose of feature extraction. Support vector machines are then employed to classify and determine whether a particular subject is a heart murmur patient. The diagnosis accuracy achieved through this approach reaches a commendable 82.7%.

Keyphrases: Cardiac Abnormality, Phonocardiogram (PCG), child, physiological data

@booklet{EasyChair:15430,
  author    = {Yuyu Yen and Chun Chang and Jui-Hung Kao and Woei-Chyn Chu},
  title     = {Development of a Classification Model for Pediatric Cardiac Abnormality Detection Through the Integration of Child Physiological Data and Phonocardiogram Features},
  howpublished = {EasyChair Preprint 15430},
  year      = {EasyChair, 2024}}