This article discusses the many challenges in performing measurements of data such as heart rate, blood glucose, blood pressure, temperature, oxygen levels, and ECG data, including the latest technologies for these measurements. The article wraps up with a discussion about batteries and suggestions to reduce power consumption in battery operated devices like wearables.
Wearable devices have become mainstream technology. This was driven by several trends
- Self-monitoring of health data
- Healthcare providers monitoring people at home
- Healthcare providers monitoring people in the hospital
People are self-monitoring to track their exercise and their healthcare providers have changed dramatically from not wanting data from wearable devices to welcoming the additional data. Doctors find the data can give them valuable insight to help diagnose their patients that they can’t get from infrequent visits by these patients.
Hospitals are highly incentivized to keep people healthy at home, since, in the United States, they are no longer paid to treat patients who return within 30 days with the same condition. Wearable devices help them monitor patents at home. Hospitals are also monitoring patients with patches that are typically applied to the chest or abdomen to collect health data instead of relying on nurses to do it periodically. The result is more frequent measurements and more accurate data.
The COVID pandemic created the need to monitor patients at home to reduce visits to the hospital. The FDA put into place rules that made wearable devices easier to get approved and used. This greatly increased the use of wearable devices.
Designing wearable devices to measure health data is not easy. There are different ways to measure each parameter, and each sensor has advantages and disadvantages. Where the sensors are located and how the data is processed is critical. Here are some common sensors and the advantages and limitations you need to be aware of when designing wearable devices.
Body Temperature
While sensing temperature is easy and the sensors are low cost, usually it is desired to measure core body temperature. Short of placing a sensor in the mouth or rectum, the measurement will be an external measurement of skin temperature. Skin temperature is often very different from core body temperature, depending upon the room temperature, the level of activity of the person, and the recent temperature exposure (such as hot or cold temperatures outdoors). Some locations on the body are better than others. Under the arm or on the forehead are very good locations, but these may not be convenient for a particular wearable device. These limitations can be overcome by using other sensors to detect the status, such as motion to know the person’s activity level or an ambient temperature sensor to know the temperature of the environment. By combining data from multiple sensors, software can determine when the temperature is likely to represent core body temperature. This is called “sensor fusion”.
Motion
While motion sensors have most often been used to measure step counts, they can do much more. They are used to monitor gait. They can be used to sense whether a person is walking, standing, or sitting. They are also used to sense falls, and they can be used for dead reckoning (position measurement) when indoors or to save power by turning off a GPS sensor. Motion sensors lose accuracy over time, so the GPS needs to be turned on periodically to get a new position.
Many of these measurements are made possible by the huge investment the motion sensor manufacturers have put into creating software. They have created code that can make these measurements when placed on different places, such as the wrist, ankle, or torso
Heart Rate
Heart rate can be measure by several technologies
- ECG electrodes
- PPG sensor
- Pressure sensor
2 ECG electrodes can measure heart rate even on extremities, such as the wrist, where a full ECG measurement does not work.
The PPG sensor, or Pulse Plethysmograph, is the sensor used to measure blood oxygen. It is even better at measuring pulse and works in locations where blood oxygen sensing is difficult.
Pressure sensing, although rarely used, can sense the pulse in the wrist or other places, just as you can feel the pressure of the pulse with your finger.
Blood Oxygen
Oxygen saturation in the blood is measured with the PPG sensor. There are two techniques:
- Transmissive PPG
- Reflective PPG
Transmissive passes two wavelengths of light through the body. One is sensitive to the Oxygen in hemoglobin, but it is also sensitive to pulse. The other is sensitive to pulse. Subtracting them gives just the Oxygen signal. It only works where the distance is short, such as the fingers or the ear. Reflective measures the reflected light from the two wavelengths, and it can be used in more locations. The amount of light reflected is small, so the signal is harder to measure with accurate results. There is still a need for good blood perfusion, so the wrist is a challenging location to make accurate measurements. It works well on the abdomen, but not over bones, such as the ribs or breastbone. It works very well in the ear, which is now becoming common, as many people are used to putting ear buds in their ears.
ECG and EEG
The “sensor” is just an electrode placed against the skin with an electronic circuit to measure the tiny electronic signal, which is in the millivolt or microvolt range. It is desired to use dry electrodes for convenience, but wet electrodes give better results. With new materials and careful design, dry electrodes are becoming commonplace.
ECG electrodes need to be well separated – at least 1.5 inches (4 cm) on the chest, and much more elsewhere. ECG cannot be measured on the wrist. The way some commercial wrist-worn devices measure ECG is by touching the device with a finger from the opposite hand. This provides a good signal with very large separation, but it stops measuring when the finger is removed from the device.
EEG can only be measured on the head. It has been successfully measured on the temples with eyeglass devices, on the forehead with head bands, and on the scalp. Successful measurements have been made with dry electrodes on the scalp without shaving the hair, but it is challenging.
Respiration Rate
The standard measurement technique for respiration is a chest strap, which works well in a shirt, but not well in many other types of devices. Another technique that works well is thoracic impedance, in which a voltage is sent into the chest, and the current is measured (or vice versa). The electrical impedance changes as the chest moves. This has been attempted from the wrist, but with the arms (higher impedance) in series with the measurement, the signal is much smaller, and I am not aware of a successful measurement that works on most people.
Blood Pressure
While an arm cuff is the standard measurement, it is not a convenient wearable device. An alternative is “pulse transit time”. Blood pressure is proportional to the change in the time delay between the heartbeat (measured with ECG electrodes) and the arrival of the pulse at an extremity (usually measured with a PPG sensor). This technique has not been medically accurate, however. Various companies have introduce technology to overcome this problem with advanced sensors and software algorithms, but the technology is covered by patents or in some cases it is not made available.
Another way that blood pressure has been measured is from the ECG signal alone, using neural network signal processing. This has only been done in the laboratory, and it has not been tested for accuracy on a large number of people.
Blood Glucose
Blood glucose today is measured with continuous glucose monitors that don’t require a finger prick. They use microneedles or sometimes a single wire that passes through the skin to get good blood contact. Microneedles (which feel like sandpaper rather than needles) are comfortable and convenient, but they require calibration with a needle stick in the finger once or twice a day. These sensors do not work on the wrist, where a lot of wearable devices are located.
Today you can combine a continuous glucose monitor and insulin pump with software to make an external pancreas that operates with closed loop control, similar to a real pancreas.
Batteries
Battery technology is a major limitation in wearable devices. People want much longer times between charging and much smaller batteries. If battery technology had improved as rapidly as semiconductors, we would today have a battery the size of the head of pin that would power your car and cost one cent. We will never get there.
Batteries are limited by chemical energy storage. Gasoline is about the highest energy density available, but it suffers from poor safety. A Lithium standard cell (not rechargeable) is about a tenth the energy density. Rechargeable Lithium cells are about half again the energy density, and they have safety issues. Alkaline cells are about half the energy density of rechargeable Lithium. With the large research effort on better batteries, I think we may see as much as a two times improvement in batteries in the next few years.
Figure 1. Energy Density of various battery types compared to gasoline. This table shows the volume energy density in a logarithmic scale.
Power
If batteries won’t improve, we need to design better electronics. The important areas to focus on are:
- Wireless transmission
- Sensors
- Displays
- Microprocessors
- Software
Wireless Transmission
There is a very wide range of power used by different wireless technologies. The power depends upon the distance the signal is sent and the rate of the data. See Figure 2.
Figure 2. Distance and speed versus power for several wireless communication standards. The numbers are power in milliwatts. BLE is Bluetooth Low Energy. The power used by 5G in the high frequency band is much greater than shown above.
Today most wearable devices use Bluetooth Low Energy (BLE), which is the most energy efficient wireless transmission widely available. They typically communicate to a smart phone and let the phone use its bigger battery to transmit data to the Internet by cellular or Wi-Fi. Bluetooth LE is limited in its data rate. It will not handle video, even with compression. Even audio is challenging. Most data on wearable devices is collected at slow data rates, so the speed is not usually an issue. ECG signals are typically the fastest, needing about 500 samples per second, which Bluetooth LE can handle. Local processing on the device can reduce the data rate.
There is a new class of wireless transmission that is very low in power, but has the range of cellular. LoRa is one. Others are SigFox, NB-IoT, and LTE-M. They are ideally suited for wearable devices, but the technology to receive the signals is not built out in all areas. Until it is, manufacturers don’t want to build devices that don’t work everywhere. NB-IoT and LTE-M, mainly at cell sites, reach much more than 90% of the US population and are widely available in most of the world, making these suitable for use in products. LoRa and SigFox lag far behind.
Sensors
Choosing sensors carefully is important for managing power. In figure 3 notice that cameras are very power hungry. GPS and PPG sensors use moderate power. Power can be reduced further by having the software turn off the sensors when they are not being used. A 9-axis motion sensor is much lower power than GPS. This is why the motion sensor is used for position sensing when the GPS is turned off.
The lowest power is the 3-axis accelerometer. They are also very low cost, less than a dollar in volume. They are often left on continuously while everything else is turned off. When they sense motion they can turn on the microprocessor, which can query various sensors to see if anything needs to be done, such as measuring steps. If the person is actually not moving, for example, everything is turned off again except for the accelerometer.
Figure 3. Power consumption of several sensor types. 10 to 20 mW is moderate. A 0.01 mW device can operate on a coin cell for years.
Displays
Color LCD (with a backlight) and color LED displays are very power hungry (see figure 4). OLED displays use less power, and they can still display video. The very low power LCD displays that you see on a wristwatch cannot display color or video, but they can run on a coin cell for years. Digital paper is a newer technology that uses almost no power until the data is changed. It is ideal for things like the Kindle reader. Digital paper cannot handle video, and its color capability is poor. To be read in dim light requires a backlight, which is not low power.
Usually the wearable device has no display, and data is displayed on a cell phone or in a web site.
Figure 4. Energy consumption of various types of displays. The horizontal scale is logarithmic. The exact power varies considerably for different models and greatly for different sizes.
Microprocessors
Microprocessor power ranges from tens of Watts to microwatts depending upon the data processing they do. A cell phone has a very powerful processor that is optimized for energy consumption, but it still requires a moderately large battery.
The processors that consume microwatts are very limited in what they can do. It takes an intermediate power level to do data compression. However, it is usually better to spend the power compressing the data than to use the power to send it uncompressed over wireless communication. Low power microprocessors have advanced power management, which means different parts of the processor can be turned off individually.
Software
As we have seen, the software is critical to saving power. It can make decisions about when to turn off power consuming devices like wireless transmitters and sensors. It can also compress data. Software can collect data and send it in bursts to improve the efficiency of wireless transmission.
About the Author
Mr. Walt Maclay, Founder and Chairman of the Board of Voler Systems, is recognized as a domain expert in Silicon Valley technical consulting associations. He has spoken at dozens of events on sensors, wearable devices, wireless communication, and low power design. In 2019 he wrote Highly Successful Engineering Design Projects, a book about project management now available from Amazon. From 2008 to 2010 he was President of the Professional and Technical Consultants Association (PATCA). He is a senior life member of the Institute of Electrical and Electronic Engineers (IEEE) and a member of the Consultants Network of Silicon Valley. He has been an instructor at Foothill College in the Product Realization Certificate Program, teaching successful new product introduction skills. He has applied his outstanding leadership to many multidisciplinary teams that have delivered quality electronic devices. Mr. Maclay holds a BSEE degree in Electrical Engineering from Syracuse University.
Mr. Maclay is active in helping technology startup companies. He has participated in angel investor groups and has advised dozens of startup companies on technical and funding issues. He has been a reviewer for NSF SBIR grants. As founder of his own company, he has dealt with issues of funding, product development, marketing, sales, and finance, giving him the experience to advise others. Voler Systems is a member of a technology consortium, the Product Realization Group, which provides all the services to introduce new technology products. Mr. Maclay started and led two forums for CEOs of small companies to discuss issues of importance to them.
About the Company
Voler Systems (www.VolerSystems.com), a dba for Strawberry Tree Incorporated, is an electronic and software design services company. It provides design, development, risk assessment, and verification of new devices for medical, consumer, and industrial applications. Voler is particularly experienced in designing wearable and IoT devices, using its skill with sensors and wireless technology. Voler Systems, one of the top electronic design firms in Silicon Valley, has highly experienced engineers who can deliver quality electronic products on time and on budget. The company has developed hundreds of products including a heart pump, augmented reality glasses, sleep monitoring devices, a golf game tracking device, and hundreds of other medical, consumer, and industrial devices.
The company has been located in Silicon Valley for over 45 years and provides design services all over the world. It has developed a large network of service providers to provide many services besides its core services of analog design, digital design, firmware, and software. It partners with other design firms to provide complete design services, including mechanical design, optical design, regulatory and quality consulting, and other services. With these high quality partners Voler can provide complete product development.