Designing smart wearable connected devices for medical monitoringFollow article
The advent of intelligent and wireless connected wearable monitoring devices is revolutionising health care. The trend started with simple fitness straps and has quickly evolved into devices that can monitor vital signs and report a change of a patient's indicators.
These smart wearables are not only useful for monitoring an individual's fitness performance but are equally valuable for the remote monitoring of patients requiring long term clinical care. Rather than needing to attend an outpatient clinic, patients wearing a health monitor can stay in the comfort of their own home. Health advice, particularly for the elderly, can highlight when an individual is not moving around sufficiently during the day.
This article investigates some of the design challenges for smart wearable IoT devices such as fitness trackers, heart rate monitors, and smartwatches. We also highlight some of the techniques and sensors used to monitor vital signs, such as heart rate, heart rate variability, and blood oxygen levels.
Advances in health monitoring
It's not that long ago that you needed to attend your local health centre or hospital outpatient's department to have your heart rate checked. The process would involve sensor patches placed on your chest that are connected to an electrocardiogram monitor. Today, chances are you are wearing a device that monitors your heart rate every minute of the day and night. Heart rate measurement is now a standard feature of most wrist-worn smartwatches, fitness trackers and health monitors. Another type of heart rate monitor more closely based on the clinical ECG monitor is the strap worn around the chest. These are still available, but for most practical purposes they are uncomfortable worn continually. For occasional use and quick clinical assessments, the finger-tip heart rate monitors provide a reasonable accuracy level.
Figure 1 - Portable heart monitor
Before getting into some of the technical challenges involved in designing a smartwatch health monitor, let’s quickly review the types of vital signs measured, how measurements are performed together with some practical design considerations.
Patients may be required to have their heart rate monitored continually for 24 hours or more for clinical diagnostic purposes. Typically, the electrocardiogram measurement (ECG) is preferred for this measurement and involves placing three or more self-adhesive electrodes to the patient's chest. The electrodes detect minute electrical impulses from the heart and determine slow, fast or irregular (arrhythmias) heartbeats. Another approach to detecting the heart rate is using optical sensors on a wrist-strap that detects blood movement through arteries under the skin.
Blood oxygen saturation level, Sp02 is also a vital sign and essential for living cells and tissues. Typically, this measurement is conducted in a similar way to the optical heart rate sensors.
Another measure of heart rate function is heart rate variability (HRV). HRV is an indicator of the heart rate's dynamic changes over time and is increasingly a feature used on smartwatches to indicate stress levels and activity. Many smartwatch manufacturers use HRV, stress levels, and exercise activity to indicate a wearer's energy level: for example, Garmin's Body Battery function.
Perhaps the most significant difference between measuring vital signs in a clinical environment and personal fitness measurements is movement. In a hospital or health centre, the clinician will take readings with the patient sat still. For a smartwatch measuring personal performance or health monitoring, the wearer will most likely be active, such as running, swimming, or kayaking.
The technical challenges associated with wearable device design
Designing a smartwatch that incorporates the ability to monitor vital signs is a significant challenge. Many considerations broadly fall into two categories: mechanical and electronics. From a marketing specification perspective, physical design and aesthetics, battery life, and useability influence development.
Firstly, the size of the smartwatch is a major limiting factor. The watch will require a significant amount of circuitry, display, power source, and sensors. It needs to fit neatly on the wrist, not be bulky or heavy, be comfortable for constant use, not snag on clothing, and be exposed to environmental extremes. Consumer preferences will heavily influence the form factor, colours and materials. The environmental considerations include the wearer immersed in water, extremes of temperature and humidity, exposure to dust and liquid; in short, all the places a human might go for necessity or adventure. Ingress protection is paramount; however, optical sensors or electrodes for vital sign monitoring and mechanical buttons to switch functions add another complication to preventing fluid or dust ingress. Touch screens are usually limited to wearables with specified use cases, such as only in a gym and cannot be worn in the shower or swimming. A charging port or wireless charging sensor also adds to the mechanical design complexities. Maintaining close contact with the wearer's skin across various use cases and activities is essential to achieve accurate results. Also, with optical sensors, ambient light rejection and eliminating direct-light paths from photodiodes - explained in the next section - is essential and requires the innovative use of mesa layers and light barriers both externally and within the smartwatch enclosure.
A final mechanical consideration relates to the use cases and whether the wearer may undertake a physically demanding and strenuous activity such as rock climbing, diving, and surfing. The likelihood that the smartwatch and its strap will get knocked or scratched will influence the material the case and strap is constructed from and the lettering methods employed around the bezel.
There are many technical considerations associated with the electronic circuitry, of which the most likely candidate, particularly from a buyer’s perspective, is the power consumption that directly relates to battery life. There is a balancing act between the physical space allocated to the battery and the space available to incorporate extra functionality. Ultra-low-power wireless connectivity is a must so that the smartwatch can pair with a smartphone, Bluetooth Low Energy (BLE) being the most preferred option. Many smartwatches also incorporate MEMS accelerometers, gyroscopes and magnetometers for other functionality. Displays typically consume high levels of power during use, so they require careful consideration. The ability to see the display in bright sunlight is essential. Many consumers may deem a colour display necessary, but it rarely adds to the useability in practice. E-paper and transflective displays have ultra-low power consumption characteristics and contribute to operation on a single charge measured in tens of days compared to days. Some smartwatches incorporate energy harvesting or small solar sensors to extend battery life during use. Many smartwatches, particularly those designed for outdoor recreation and adventure, will incorporate a GPS receiver rather than utilise constant pairing to a smartphone's GPS receiver. GPS receivers have a relatively high power consumption profile, so battery life is considerably reduced during use.
Using sensors to detect heart rate and oxygen levels
Wrist-mounted smartwatches use a combination of LEDs and photodiodes to detect a heart rate under the surface of the wearer's skin.
Figure 2 - Basic principle of optical heart rate detection using LEDs and photodiodes (source Maxim Integrated)
Heart rate measurement
The skin consists of seven main layers, and each layer exhibits different thickness, absorption and scattering coefficients. The layer of interest for the pulsing heart rate detection is in the sixth layer from the surface, the inferior, or lower, blood net dermis. Detecting a pulsing heart signal from the blood flows in this layer is termed photoplethysmogram (PPG). For the optimal detection of the most prominent PPG signal, the LED wavelength needs to be within the absorption peaks of the blood's haemoglobin (Hb) and deoxyhaemoglobin (Hb02), these occurring between 540 to 570 nm. The closest wavelength that commercially available LEDs have are green LEDs operating at approximately 530 nm. The photodiodes used for sensing the reflected light also need to be optimised for operating at this wavelength. The blood pulses of the PPG heart come from the regular contractions of the heart. As mentioned above in the mechanical considerations section, the photodiodes and the LEDs must be optically isolated. Typically, this is achieved using a raised mesa underneath the smartwatch with the LEDs and photodiodes surrounded by optical separation techniques. See Figure 2. This approach also assists in reducing the impact of ambient light sources on the PPG detection process. Unwanted ambient light sources can include direct sunlight, street lighting, fitness mood lighting, and the internal LCD or OLED display.
Figure 3 - Physical construction of LEDs and photodiodes on an example smartwatch (source Maxim Integrated)
However, the PPG signal also includes artefacts caused by the wearer walking, jogging, or simply shaking hands. The wrist motion creates these unwanted signals, but also the wrist movement also changes the separation between the smartwatch and the skin, resulting in an air gap. To determine the correct heart rate signal requires removing the motion artefacts imposed on the PPG signal, which is typically achieved by using an accelerometer. Signal processing of the photodiode output and the accelerometer outputs involves detecting the motion frequencies and eliminating them from the raw PPG signal. While the accelerometer yields motion outputs, they only tend to be useful for regular motion movements such as running or walking at a relatively constant speed. Another approach, often combined with an accelerometer, introduces multiple optical paths by using two photodiodes. The accelerometer output and the two PPG signals are then processed using techniques including a Kalman filter, motion compensation algorithms, and an adaptive notch filter to remove the unwanted motion and noise artefacts from the PPG outputs.
The measurement of the amount of oxygen, Sp02, in the blood is a vital indicator of a person's general health. Oxygenated haemoglobin (Hb02) and deoxygenated (RHb)are the two key types of haemoglobin in red blood cells. The Sp02 is a percentage measure of Hb02 compared to the total and requires measurement of Hb02 and RHb values. These measurements are performed similarly to the heart rate by using LEDs and photodiodes. However, because the absorption wavelengths of Hb02 and RHb are different, two LEDs are used. Typically, the LEDs are a red LED operating at 660 nm, and an infrared LED operating at 880 nm. Unlike heart rate measurement, Sp02 measurement typically requires calibration and compliance testing against the international standards ISO 80601-2-61:2017.
Smart IoT medical devices advance health monitoring
In this article, we have investigated the wearable devices such as smartwatches have revolutionised the way health monitoring of our vital signs is achieved. Smartwatches provide a convenient and comfortable way to monitor key vital signs continuously, whether you are a patient recovering at home or a keep-fit enthusiast who likes to be in touch with their general health and fitness levels. Heart rate and pulse oximetry readings can be taken by anyone, at any time, and, through an attached smartphone app, recorded.
No longer is it necessary to attend a health centre or clinic to have vital signs checked. If the smartwatch detects abnormal vital signs, the wearer can arrange a full clinical assessment based on what the smartwatch may have highlighted.
Biological sensor technologies continue to evolve, so very soon we can expect new smartwatch features that can measure our blood pressure and chemical metabolism indicators such as glucose and lactate.