Listening to Your Heart

Wearable devices, such as smartwatches and wristbands, have forever changed the way individuals can track and manage their health. These devices have evolved from simple step counters to sophisticated tools capable of monitoring various physiological parameters. They are equipped with sensors that can track heart rate, blood pressure, sleep patterns, and even oxygen levels. By continuously collecting data, they provide wearers with valuable insights into their overall well-being. This real-time monitoring allows users to make informed decisions about their lifestyle, exercise routines, and overall health management.

One of the most important benefits of wearable health devices is their ability to detect medical conditions early. By continuously monitoring physiological parameters, they can help identify irregularities or abnormalities, prompting users to seek timely medical attention. For example, sudden fluctuations in heart rate or blood pressure beyond normal ranges can indicate potential cardiovascular problems. Similarly, irregular sleep patterns can be a sign of underlying sleep disorders. Early detection of these conditions can result in better health outcomes and an improved quality of life for individuals.

Despite their promising capabilities, the adoption of wearable health devices remains limited. Cost, inconvenience, and a lack of awareness about their benefits are among the key factors hindering widespread acceptance. In contrast, the use of headphones and earbuds has surged significantly in recent years, becoming ubiquitous in daily life. However, their potential for health monitoring has largely remained untapped.

Engineers at Google recently teamed up with researchers at the University of Pittsburgh and Rutgers University to explore the health monitoring potential of hearable devices to bring these benefits to a wider audience. In particular, they investigated how they could leverage the components of an active noise cancellation system, as this technology has become very common even in lower-end devices recently. They realized that the paired speaker and microphone available in these systems could be used for audioplethysmography — a technique that can be used to measure volumetric changes in different parts of the body.

The team used this technique to alter the operation of commercial headphones such that they would actively emit inaudible, ultrasonic probing signals into the ear canal using their speakers. These signals interact with the ear canal, then are reflected back in the direction from which they came, where they are received by the microphones that are normally used by a noise cancellation system to sense sounds heard by the user.

These ultrasonic signals are modulated by their interactions with the ear canal, and the ways in which they are changed can be used to determine the shape of, and changes in the shape of, the ear canal. This is important, because cardiac activity causes deformations in the surface of the ear canal as blood flows through nearby vessels. As such, the signals can be leveraged to infer physiological parameters like heart rate and blood pressure.

The signals produced by these deformations are incredibly tiny, and are undetectable under normal circumstances. Accordingly, the researchers adopted a technique from the wireless communications field called coherent detection. This involves mixing the weak received signal with a strong oscillating wave, then extracting the true signal by subtracting the oscillating wave from the received signal. This enabled the team to develop a system that can operate at a very low intensity, which is well below the generally accepted safe limits, and that can also produce accurate measurements even when music is being played by the headphones at the same time.

An eight-month long study of 153 participants was conducted to assess the performance of the headphones. It was found that they could detect heart rate and heart rate variability with a high degree of accuracy, with 3.21% and 2.7% median error rates, respectively.

Moving forward, the researchers plan to explore ways in which they can better deal with noise that can enter the system as the user is in motion (e.g. hiking, weight lifting). With refinements such as these, they believe their system will one day be capable of measuring many more physiological parameters.