Cheap Wi-fi chips like those found in a $30 Raspberry Pi can measure the human pulse with the accuracy of a clinical heart rate monitor or an expensive health and fitness tracking device like the Apple Watch.
According to UCSC researchers who headlined the Pulse-Fi study, a simple Wi-fi network created with a five-buck ESP32 chip can track one's heart rate no worse than the Apple Watch 10 that is currently discounted on Amazon but still carries a $359 price.
The test results from the cheapo Raspberry Pi were even more accurate, as the researchers sifted Wi-Fi Channel State Information (CSI) data through AI algorithms to detect the heart rate of more than a hundred study participants.
The characteristics of the Wi-Fi channel created between a transmitter and a receiver, such as phase, frequency in physical surroundings, or amplitude, can change slightly with each breath and heart beat. Those miniscule changes are then filtered with the help of machine learning algorithms that discard all other causes that can change the CSI of a Wi-Fi network, leaving the Raspberry Pi with the correct pulse measurement of all the 118 people involved in the study.
To top it all off, the Wi-Fi network heart rate detection abilities were there regardless of the pose the participants were in, or whether they were moving, standing, sitting, and even lying down.
To achieve this, the team had to develop their own database from scratch and use a control device like a clinical-grade oximeter to teach the AI algorithms which changes in the Wi-Fi channel frequency or amplitude could be attributed to a heart beat, and which belonged to interference from other sources.
The AI-driven set they deployed helped detect pulse from a longer distance, paving the way for casual heart rate monitoring by Wi-Fi networks with the Pulse-Fi algorithm. Beside Wi-Fi pulse detection, the UCSC researchers are now targeting breathing rate pattern recognition, which can help people living with sleep apnea, too.