Tuesday 25 February 2025
A team of researchers has developed a novel system for detecting abnormal heart sounds using mobile phones and a lightweight neural network optimized for on-device inference. The system, which doesn’t require any additional equipment or specialized training, has the potential to democratize early diagnosis and treatment of cardiovascular diseases.
The traditional method for detecting heart murmurs involves medical practitioners using stethoscopes to investigate irregular sounds, followed by echocardiography and electrocardiography tests. However, these methods can be time-consuming and may not always provide accurate results. The new system uses a mobile phone’s built-in microphone to record the heartbeat sound, which is then analyzed by a neural network designed specifically for this task.
The researchers developed an Interpretable Convolutional Neural Network (IConNet) that harnesses insights from audio signal processing to enhance efficiency and transparency in neural pattern extraction from raw waveform signals. Unlike previous approaches that rely on specialized stethoscopes or complex preprocessing pipelines, IConNet is optimized for on-device inference, making it suitable for real-world applications where recording environments and audio quality can vary.
The system was evaluated using a dataset of 2575 normal and 665 abnormal heart sound recordings, with the results showing an accuracy rate of 92.05%. While this is not yet at the level of state-of-the-art models, it demonstrates the effectiveness of the proposed method in classifying heart sound data.
One of the key advantages of the system is its ability to learn features from raw audio signals without requiring any preprocessing or feature extraction steps. This eliminates the need for manual annotation and reduces the risk of human error. Additionally, the neural network’s interpretability allows researchers to understand which features are being used to make predictions, ensuring that the results are trustworthy.
The potential impact of this technology is significant, particularly in low-resource settings where access to medical equipment and expertise may be limited. By enabling early detection and treatment of cardiovascular diseases, the system could help reduce the global burden of these conditions and improve healthcare outcomes.
In the future, the researchers plan to integrate data from wearable trackers and other sensors to further streamline heart health monitoring and provide more personalized recommendations for medical check-ups. As the technology continues to evolve, it has the potential to empower individuals to take a proactive role in managing their own health and well-being.
Cite this article: “Mobile Phone-Based System for Detecting Abnormal Heart Sounds”, The Science Archive, 2025.
Mobile Phones, Heart Sounds, Neural Network, Cardiovascular Diseases, Stethoscopes, Echocardiography, Electrocardiography, Machine Learning, Audio Signal Processing, Healthcare Outcomes
