Phonocardiogram Classification Using 1-Dimensional Inception Time Convolutional Neural Networks

Bjørn-Jostein Singstad1, Lars Bongo2, Markus Johnsen3, Johan Ravn3, Antony Gitau4, Henrik Schirmer5
1Simula Research Laboratory, 2The Arctic University of Norway, 3Medsensio, 4Kenyatta University, 5Department of Cardiology, Akershus University Hospital


Murmurs are sounds caused by turbulent blood flow that is often the first sign of structural heart disease in patients. These sounds are detected by auscultating the heart using a stethoscope, or more recently by a phonocardiogram (PCG). We aim to identify the presence, absence, or unclear cases of murmurs, as well as predict normal or abnormal clinical outcomes from PCG recordings using machine learning.

We trained and tested two 1-dimensional convolutional neural networks (CNN) on a PCG data set from a pediatric population of 1568 individuals. One model predicted murmurs, while the other model predicted clinical outcomes. Both models were trained to give recording-wise predictions, while the final predictions were given for every patient (patient-wise predictions).

This paper describes our participation in the George B. Moody PhysioNet Challenge 2022. The objective of this challenge was to identify heart murmurs and clinical outcomes from Phonocardiogram recordings. Our team, Simulab, trained a clinical outcome classifier that achieved a challenge cost score of 8720 (ranked 1st out of 305 submissions) and the murmur classifier achieved a weighted accuracy of 0.585 (ranked 182nd out of 305 submissions) on the validation set.