Putting a Patch on Poor Diagnostics

Genetic tests, MRI scans, AI-powered biomarker analyses — today’s physician’s office looks very little like it did 200 years ago. Yet it seems that some things never change. Consider the stethoscope, for instance. These instruments were first put to use way back in 1816, and they are still just as important today. By using a stethoscope to listen to the sounds made by the heart, lungs, and intestines, physicians can diagnose a wide range of medical conditions.

But while a stethoscope is very simple to use, it takes a lot of experience to use it well. Due to factors like external noise or the mixing of sounds from multiple organs, diagnosis accuracy can suffer. Past research has shown that medical students can only diagnose conditions such as asthma and COPD correctly about 60% of the time with a stethoscope. Even more seasoned medical professionals only get it right about 80% of the time.

Digital stethoscopes, on the other hand, are far more accurate than humans. Using a variety of algorithms to reduce noise, boost the signal, and otherwise process the audio, digital stethoscopes achieve a much better level of diagnostic accuracy. However, these devices are inconvenient to use, so they are rarely seen in clinical settings. Despite their limitations, the convenience of traditional stethoscopes has ensured that they remain hanging around every doctor’s neck to this day.

A group led by researchers at the Korea Institute of Science and Technology believes that the time is right for this to change. Taking advantage of the rise of wearable electronics, the team has developed a wearable digital stethoscope called the Lung-Sound-Monitoring-Patch (LSMP). This combination provides all the accuracy of a digital stethoscope with the convenience of a wearable device.

The new system is targeted at monitoring lung sounds, and as such, it attaches to the skin over the chest. Inside of LSMP is a uni- and omni-directional MEMS microphone, a Nordic Semiconductor nRF52832 microcontroller, and a LiPo battery for power. The hardware is enclosed in a 3D-printed case.

Sound collected by the microphone is analyzed with a machine learning algorithm. This convolutional neural network was trained to recognize both the sound of normal breathing and wheezing. A long-term clinical trial with a COPD patient revealed that the system is capable of detecting wheezing events with better than 80% accuracy, which puts it on par with an experienced physician.

Further experiments demonstrated that the device, despite being designed for analyzing lung sounds, could also measure heart rate and respiratory rate. LSMP outperformed existing digital stethoscopes in terms of accuracy in the process.

Looking ahead, the researchers envision applications in telemedicine, home health monitoring, and even emergency response — anywhere that quick and accurate diagnostics are needed. As wearable electronics continue to become more advanced and comfortable, devices like this could help bring the power of modern diagnostics to more people, more easily.