Designing affordable medical sensors: from concept to test to market
01 October 2021
LMD_e-Checkup Sensor_on smartphone
Medical technology has historically been trapped in a vicious circle. Development & manufacture are costly – not least because of strict regulatory controls – so devices are expensive. Not many are sold, so there are few economies of scale. For medical devices, it follows therefore that there is no viable middle ground: either cost spirals up & sales spiral down – or the opposite occurs, resulting in a virtuous circle.
This article was originally featured in the October 2021 issue of EPDT magazine [read the digital issue]. And sign up to receive your own copy each month.
Here, Dr Chris Elliott FREng, founder of medtech pioneer, Leman Micro Devices (LMD) explains LMD’s six rules for achieving the latter outcome…
LMD was founded almost a decade ago with a mission to develop medtech devices that would break the vicious circle and move to the virtuous one. Health gains in wider populations can only be achieved by developing devices that achieve mass market adoption, so we set out to develop products that deliver true health benefits which can be manufactured at scale to amortise their regulatory costs. As a result, they would be inexpensive and everyone could afford them.
Here are our six rules to achieve this:
Rule 1: Market drive
The world is full of great ideas for medical technology – but ideas are where many of them remain. When we founded LMD, we spent the first six months looking at large-scale health problems and whether technology could be developed to solve some of them – and then working out what would sell in massive quantities if we could make it and, crucially, how to take it to market. Only then did we start thinking about how we would achieve it. We targeted over-the-counter sales to consumers; accessing patients requires not only that your idea is good for their health, but also that it appeals to people who take buying decisions – hospital administrators, insurance companies and reimbursement managers. A great idea needs even greater marketing; technology alone is not enough. The phrase, “Build a better mousetrap, and the world will beat a path to your door”, attributed to American writer and philosopher, Ralph Waldo Emmerson, is simply wrong!
Rule 2: Use new technology, avoid new science
There are hundreds of thousands of clever scientists around the world, inventing new diagnostic and sensing science. To get to market, it needs to be expressed as effective, reliable, useable and affordable technology. The opportunity for engineers and technologists is to apply miniature and inexpensive technologies to that science.
Figure 1. Vicious and virtuous circles
I can illustrate with LMD’s experience. When we saw the global data for deaths from hypertension (which is a manageable disease when someone knows they have it), we decided that measuring blood pressure would be critical for our device. We studied hundreds of scientific papers proposing new physical principles based on the propagation and reflection of pressure waves in the arteries. None of those had managed to measure blood pressure (and still haven’t), so we adopted the tried and tested principle of the cuff (balance the pressure in the cuff against the pressure in the artery), but brought it into the 21st century with miniature technology. As a result, LMD’s team of mechanical and electronic designers, software developers and medical advisors developed the only sensor that can measure absolute blood pressure – and yet is small enough and cheap enough to build into a smartphone. Despite being based on science more than 100 years old, the V-Sensor is recognised as innovative, with many patents granted around the world.
Rule 3: Minimise the smarts
Most of us walk round with a 2GHz computer in our pockets – and we have an amazing thinking engine in our skulls! We drove down costs by keeping the sensor as simple as possible, with our custom ASIC to capture the analogue data from the pressure, optical and temperature sensors, and an analogue-to-digital convertor (ADC) connected to an I²C bus. It does no processing or analysis; we leave that to the phone or tablet at the other end of the I²C.
This drives down cost: using the computing power in the phone and the judgment in the user, we can eliminate most of the expensive electronics and mechanics.
Rule 4: Design for manufacture
You don’t create an inexpensive sensor by taking an expensive one and making it cheaper: you must design from the start for large quantities and low cost. That means avoiding any intrinsically expensive components or technology, and always thinking about how it will be assembled and tested.
We chose to have our early experimental devices built by a partner with the capability to manufacture hundreds of millions of units per year. That introduced extra development costs and management challenges, over the simpler route of building them by hand in a clean room, but it ensured that we were always working with a design that could be produced in quantity.
Rule 5: Design for test
Few MEMS devices are manufactured to a high enough tolerance to achieve medical accuracy, so it is necessary to calibrate each one. For many sensors, the cost of test and calibration is as much as the cost of components.
Figure 2. The V-sensor
LMD’s calibration facility was designed at the same time as the V-Sensor, and we included a critical feature in the sensor’s ASIC. On probe test, every good die is given a unique identity number (64 bits – we have big ambitions!) that can be read via the I²C bus. After that die is built into a sensor, we can track it and its history.
This enables a streamlined system for 100% test and calibration, and for distribution. We developed two test stations: a large-area black body to calibrate the thermopile, and a chamber with controlled pressure and temperature to calibrate the pressure measurements. 256 sensors are placed by a robot gripper in sockets on a test board with multiplexed I²C, so each can be read, together with its unique identity number. One board at a time goes under the black body, four at a time go into the chamber. With a 90-minute cycle, we can calibrate over 5 million sensors per year, and the chambers can be replicated as needed.
Calibrated devices are sent as a batch to the assemblers who incorporate them into host products, like a smartphone. There is no need to track each one because, when the user first activates the sensor, it connects to our servers and downloads its calibration parameters.
There’s another elegant consequence of this approach. The robot gripper does more than manipulate the sensors, it connects to them and, during the two seconds it takes to place each on the board, it performs a full test of functions. It also uses a spectrometer to measure the central wavelength of the LEDs. Instead of buying expensive LEDs with a known wavelength, we can use LEDs that emit with ±30 micron and then use an equation to correct for their actual wavelength – that takes several cents out of the BoM (bill of materials).
Rule 6: Design for use
Finally, there’s no point in an accurate and affordable device if no-one can use it. From the start, we considered how people would be able to use our device to make reliable measurements. That includes the ergonomics, as well as the instructions for use, feedback to help users learn, and simply ensuring that it is all in line with their language and cultural expectations.
By following these six rules, we’ve developed a sensor that many consumer product companies – large and small – are lining up to buy, just as soon as we ramp up the production rate. That’s exciting for us, but it also points to a bigger picture.
Figure 3. The test chamber
We are at the start of a revolution in medical sensors. The first generation were toys: they measured outputs (like number of steps), rather than inputs (your level of health). The second generation started to measure what could be useful (like pulse rate), but were still limited to easy targets. The third generation are regulated medical devices, that make medically-accurate measurements of clinically-significant parameters (such as blood pressure or temperature), and are small enough and cheap enough to be “always with me” (built into smartphones or wearables, for instance). As they integrate with AI diagnosis, we will see a revolution.
How it works: the V-Sensor
The V-Sensor has a MEMS (micro-electromechanical systems) pressure sensor embedded in flexible resin that transmits the pressure from an index finger placed on it to the ASIC (application-specific integrated circuit). That’s a key difference from all the other approaches – we measure blood pressure with a pressure sensor.
Most people are unaware that there are arteries in the human index finger which are close to the surface of the skin. Pressure on those arteries cause them to collapse or occlude in the same way that a pressure cuff does when applied to the upper arm. A game-style interface on the smartphone app ensures that the user holds the correct finger pressure for around 45 seconds.
The V-Sensor also has an optical sensor, consisting of two LEDs and a photodiode. The light shining through the skin is partly absorbed by the blood. That absorption tells us the size of the artery, blood oxygen level and the pulse rate (bpm). Respiration rate is found from the modulation of the pulse.
The V-Sensor also incorporates an infra-red thermometer that gives medically-accurate temperature readings from scanning the forehead. The ASIC also includes circuitry to detect an ECG between the two hands.
Our custom ASIC pulls it all together.
Contact Details and Archive...