Pre-screening for heart conditions is particularly challenging in low- and middle-income countries due to the lack of expensive equipment and a shortage in medically trained professionals. As heart sounds can be captured easily by smartphones or similar devices, their automated analysis may provide a cost-efficient alleviation of this problem. One potential symptom for cardiac diseases that can be detected through heart sound analysis are so-called heart murmurs.
In this study, we present an approach for detecting heart murmurs that utilizes a Pooling-based Artificial Neural Network (PANN) structure for extracting features from audio waveforms of arbitrary lengths. It can classify single recordings based on recording location and the extracted features in an end-to-end manner. The approach is inspired by the multiple instance learning framework.
We performed a 10-fold stratified cross-validation and report the calculated evaluation measures as average (standard deviation): Murmur weighted accuracy 0.715 (0.077), Outcome-Metric 13640 (2401). The official murmur weighted accuracy and outcome validation score were 0.720, 9135 respectively.