Valencell is a strong proponent of the fact that the best heart rate sensors on the market are developed with a systems approach. What exactly do we mean by systems approach? A systems approach is one that recognizes the hardware, software, testing & integration of heart rate sensors technology must work together be designed and implemented to work together to achieve high performance. It doesn’t work to have the hardware designed by one company and the software algorithms and assessments designed by another company. It’s like trying to put together a car with a Ford engine, a Chevy transmission, and a Tesla chassis. It might look like a car, but it won’t perform very well.
And these are complex systems with numerous elements involved, including hardware, software, testing & validation, and manufacturing. Here’s a look at the key components:
Analog Front End (AFEs)
The AFE is the part of the system that takes the electrical impulses from the photodiode and does the filtering, signal conditioning, and analog-to-digital conversion (among other things). AFE configuration for biometric sensing requires system level knowledge for fine-tuning to best meet the system requirements, including optical, electrical, mechanical, and algorithm requirements. For example, it’s common to find electrical noise in the sensor systems that cause inaccuracies in the heart rate sensor that may otherwise remain a mystery.
Light is both a friend and an enemy of optical heart rate monitors. Ideally the device would capture only the light from the emitters that has interacted with the capillary beds and tissue, but wearables by default are exposed to sunlight, fluorescent light, and other optical noise. You need to understand how to deal with sunlight and room light rejection, not only from a silicon configuration, but from an optical design perspective as well. This technology is advanced by optical engineers who do predictive simulations of physiological and sensor models, experimenting and exploring the ranges of how the light interacts with the human body. This is valuable for both R&D on new sensors and metrics, but also to give design guidance during the product development process.
In addition to optics, you need to understand how light interacts with tissue, blood, and skin, particularly varying skin tones, because different skin tones absorb light in different ways. Another component of human physiology at play is how motion of the body during different activities governs ideal sensor placement and optomechanical design choices. For example, periodic motion like jogging has very different physiological characteristics when it comes to optical biometric sensors than non-periodic motion like CrossFit or high-intensity interval training.
Just like industrial design (ID) has to be negotiated with electrical engineering, the ID needs to consider the biometric sensor system as well. First and foremost, sensor stability and placement on the body are critically important to accuracy. You need to have good coupling of the sensor and the skin and the gross displacement during motion must be kept to an absolute minimum. A good way to ensure this during the product development process is to have experienced mechanical engineers on the project involved in the user testing of early prototypes. They can provide invaluable feedback to guide design changes through the testing process that improves performance, comfort, and user experience. This can save time and reduce overall project risk.
Firmware & Algorithms
The firmware manages the process from raw sensor data (both accelerometer and photodiode), passing that data through the algorithm, getting output and communicating that output. If you have to reconfigure the firmware and algorithms for new hardware, it’s an entire trial and error process all over again, which costs you time and money in the product development process. The same thing applies to different use cases – is the algorithm flexible enough to support high-intensity interval training (HIIT) and lifestyle activities like typing or emptying a dish washer? Something else to consider is whether or not the device is firmware upgradable for future enhancements.
If you don’t get the hardware and firmware solid, no matter how good the algorithms are, it will result in noisy sensor data and it will suffer poor results. It’s very much a “garbage in, garbage out” scenario.
The first phase of wearables growth was driven primarily by activity tracking – counting steps, calories, and general activity levels. The first phase is over, and the next phase of wearables growth is being driven by compelling user experiences that provide deeper insights into how the wearer’s body responds to whatever activity it is doing. We’ve written about this extensively, so we won’t belabor the point, but we encourage you to read more on the next phase of wearables growth.
Biometric assessments bridge the gap between heart rate, R-R interval (RRi), activity and the interesting, engaging, personalized user experiences customers want. This process requires assessments that start with accurate sensing, transition through well-known, validated assessment methods, and end with an actionable user experience. Users of wearable devices expect more these days, and assessments are where rich, accurate data is turned into valuable, actionable information to help them accomplish a goal.
A sensor technology system can’t be judged in a conference room or by an engineer on a bench with a finger on a sensor module. It can only be judged by in-session activity testing under controlled protocols that mimic real-life use cases. In addition, testing must be performed on statistically significant number of users prior to product launch, including people of all shapes, sizes, fitness levels, and skin tones. Understanding the test “corner cases” helps guide the scope of the marketing claims that are made and helps companies avoid liability concerns from excessive promises.
Building biometric wearables at scale is very different from other consumer electronics or wearable devices, because of the optical sensor systems. They require specific test fixtures and processes for production that must match the design and hardware being produced. And just because something works in the lab, doesn’t necessarily mean it will work off the production line. It important to not lose sight of the goal so close to the finish line – engaging user experiences that aren’t offset by rashes, “sock-drawer” phenomenon, inaccuracy, poor build quality, etc. You only get one chance to make a first impression and biometric wearable.
Despite being a relatively new industry, the wearables space already has a fairly extensive IP landscape. The landscape covers sensor systems, wireless systems, processors, power, user experience and much more. Here’s a high-level view of the landscape through 2016 and it has only grown since then.
Make sure you understand what you are developing and where IP already exists before you get too far down the path.
If you are interested in more on this topic, we recently conducted a webinar on this topic, which you can view here. And if that’s not enough for you, feel free to reach out to us at email@example.com and we’d be happy to continue the discussion.