COVID-19 continues to dominate the headlines and one of the most visibly affected industries is professional sports. When basketball returned to action in isolation in July, the NBA and WNBA offered players the option of using the Oura Ring to monitor their health. The smart ring can, according to its maker, predict the onset of COVID-19 symptoms up to three days in advance with 90% accuracy.
The ring’s software platform, developed by West Virginia University’s Rockefeller Neuroscience Institute, uses artificial intelligence-based models to forecast when a player might begin feeling ill based on physiological data collected by the sensors embedded in the ring and stored on a smartphone app. Specifically, the ring collects blood pulse volume data that is used to determine heart rate, heart rate variability, and respiratory rate. The rings also collect body temperature data, hand movement data, and a time stamp. This data is then downloaded via the smartphone app to the AI platform for analysis.
To collect this information, in addition to a body temperature sensor and an accelerometer, the Oura Ring includes a photoplethysmography (PPG) sensor, a type of sensor that has been used in medical devices and wrist wearables for years. Depending upon the location on the body a PPG sensor is designed to be placed (e.g. forehead, wrist, earlobe, etc.), the sensor includes either one or more red/infrared LEDs, green LEDs, or sometimes both, as well as a photodetector for measuring the intensity of applicable wavelengths. The light intensity detected by the photodiode is then used to determine physiological parameters such as heartrate and blood volume.
IR LEDs are able to penetrate deeper into the body, e.g., into muscle tissue, but are more susceptible to motion artifacts like movement of the device over the skin, unevenness of the skin, and ambient temperature, making green LEDs a better option for some applications. Inclusion of an accelerometer to acquire direction of motion reduces the impact of motion artifacts, making the use of an IR LED a feasible option in applications where it might not be otherwise. This is the approach the makers of the Oura ring appear to have taken.
As the diagram below shows, a PPG sensor emits light that is reflected by (or sometimes transmitted through) tissue into a photodiode.
This diagram shows PPG sensor function.
Although the configuration is application-dependent, a typical PPG sensor might combine one or more LED emitters, an optical detector, and signal-processing circuitry. Optical pulsing combined with synchronous detection can reduce operating power requirements and improve rejection of ambient light.
As an example, the ADPD144RI from Analog Devices, makes use of two red (660 nm) and two IR (880 nm) LEDs and a four-segment photodetector optimized for red and IR emissions. The LEDs emit pulsed light in synchronization with an analog signal-conditioning circuit, which includes transimpedance amplification, ambient light rejection, and gain. Each of the four conditioned signals is routed to an ADC and then to an accumulator. After averaging, the resulting signal can be read either via output registers or a FIFO buffer. See a block diagram for the module below.
The operating temperature specification for the ADPD144RI is -40 °C to +85 °C. LED junction temperature is specified at 105 °C. The Human Body Model ESD rating is 3000 kV. These specs indicate a robust design that should operate reliably in indoor medical applications.
PPG technology was first explored in the 1930s and today there are a number of medical devices and wearables on the market. According to a forecast from Medgadget dated April 4, 2020, the global medical wearable market was valued at nearly $13B in 2019 and is expected to reach almost $38B by 2025.
Wearables like the Oura Ring are not, however, a panacea. There are some concerns about product efficacy, the most significant of which is the small number of studies conducted to date of the predictive ability of wearable devices in general. Other intrinsic shortcomings include measurement accuracy, mostly having to do with variations introduced by motion, which would affect the AI model’s predictive ability, as well as the timing of COVID-19 symptom onset vs. peak transmission, which may not coincide. The FDA has not to date approved wearable devices for the purpose of COVID-19 sensing.
>> This article was originally published on our sister site, EDN.
|Yoelit Hiebert has worked in the field of LED lighting for over 10 years and has experience in both the manufacturing and end-user sides of the industry.|