A new study published in Engineering presents a breakthrough in medical technology with the development of a wearable stethoscope that can accurately monitor lung sounds in real-time and automatically detect wheezing. This innovation aims to overcome the limitations of traditional auscultation methods and provide more effective respiratory disease monitoring.

The traditional stethoscope, invented in 1816, remains a crucial tool in medical diagnosis. However, its accuracy is limited by factors such as the user’s experience and the presence of external noise. Digital stethoscopes have improved in some aspects but are still not widely adopted due to inconvenience. Wearable electronics offer a new solution, and this study focuses on developing a flexible, skin-attachable stethoscope for continuous monitoring.

The key component of this new device is the Lung–Sound–Monitoring–Patch (LSMP). The main components include uni- and omni-direction MEMS microphone, microcontroller unit (MCU), lithium polymer battery. The LSMP is designed with an acoustic path to enhance sound signal acquisition. It is encapsulated in a 3D-printed biocompatible resin enclosure and attached to the skin using medical-grade adhesive.

To evaluate the performance of the LSMP, the researchers conducted a series of experiments. They tested the device’s ability to distinguish between normal and abnormal breathing sounds in healthy subjects, pediatric asthma patients, and elderly chronic obstructive pulmonary disease (COPD) patients. In normal subjects, the LSMP could accurately extract heart rate (HR) and respiratory rate (RR) from bioacoustic signals, outperforming e-stethoscopes.

For pediatric asthma patients, the LSMP was able to analyze the acoustic characteristics of wheezing and normal breathing. In one case, it detected distinct wheezing signatures during abnormal breathing periods. For elderly COPD patients, although the noisy environment posed challenges, the LSMP could still distinguish between normal and abnormal breathing through techniques like discrete wavelet transform (DWT) and continuous wavelet transform (CWT).

An AI-based algorithm was also developed for the LSMP. Using a two-dimensional convolutional neural network (CNN), the algorithm can classify breathing sounds and count wheezing events. The researchers trained the model with augmented lung-sound data. In a long-term clinical trial with a COPD patient, the AI-based event-counting algorithm achieved an 80.5% match rate with the clinician’s count, demonstrating its high accuracy in monitoring lung sounds over time.

The researchers believe that this wearable stethoscope has great potential in clinical applications. Future work will focus on integrating active-noise cancellation technology to further improve the device’s performance during daily activities and enable 24-hour monitoring. This could provide more comprehensive data for medical decision-making and help in better understanding the relationship between lung sounds and various factors such as environmental changes and drug administration.

The paper “A Wearable Stethoscope for Accurate Real-Time Lung Sound Monitoring and Automatic Wheezing Detection Based on an AI Algorithm,” authored by Kyoung-Ryul Lee, Taewi Kim, Sunghoon Im, Yi Jae Lee, Seongeun Jeong, Hanho Shin, Hana Cho, Sang-Heon Park, Minho Kim, Jin Goo Lee, Dohyeong Kim, Gil-Soon Choi, Daeshik Kang, SungChul Seo, Soo Hyun Lee. Full text of the open access paper: https://doi.org/10.1016/j.eng.2024.12.031.