Apple is focusing on health monitoring using wearable devices such as the Apple Watch. This time, they published a research paper on estimating the number of breaths using AirPods.
The number of respirations is one inhalation and one exhalation, and it is known that the normal value is 12 to 20 breaths per minute for adults. An increase in the number of respirations is a precursor to a rapidly changing condition and is also a vital sign that leads to early detection of complications.
The paper is titled Breathing Rate Estimation from Breathing Audio obtained through a wearable microphone published on the Machine Learning Research page of Apple’s official site. Although the product name of AirPods is not specifically listed, the wearable devices and headphones with a built-in microphone here can be seen as the AirPods series.
The purpose of this study is to prove that an aesthetically pleasing and inexpensive device can be used to estimate respiratory rate and track cardiorespiratory function. In addition, the paper revealed that a machine learning network model was used to distinguish between normal breathing and deep breathing based on breath sounds captured by a worn microphone. It also detects breathing sound patterns to estimate the number of breaths.
Why would you want to use a headphone microphone to measure your breathing rate? In the paper, a thermistor, a device that detects breathing conditions with temperature sensors installed on the nose and lips, a breathing transducer that measures elasticity by respiration with a belt wrapped around the chest and abdomen, and an acoustic sensor can accurately estimate the breathing pattern, but It is not comfortable for everyday use. On the other hand, headphones are relatively economical, easy to use, comfortable, and have excellent aesthetics.
While the study focused on estimating the number of breaths during physical activity, it noted that similar techniques could also be used in clinical scenarios of shortness of breath or dyspnea. Dyspnea during exercise could be a strong independent predictor of mortality, frequently used in medical research.
In this data collection, participants are prompted to record before, during, and after exercise. In addition, the heart rate measurement by the Apple Watch is also included as accompanying data, trying to capture possible data. These data are interpreted with the help of a neural network, and the results of Concordance Correlation Coefficient (CCC) 0.76 and Mean Squared Error (MSE) 0.2 are obtained, which describes the judgment that the estimation of the respiratory rate is feasible.
It is not known whether Apple will include a breathing rate search function using AirPods in iOS, etc. based on the results of this investigation. However, Apple executives have suggested the possibility of integrating health-related functions with data obtained from the AirPods sensor, and there is information that the next-generation AirPods will incorporate ALS as part of the biometric function. Research results can be used in the form of a breathing rate measuring app or an increase in health indicators. Related information can be found here.