Everyone starts their day with a routine, and like most people these days, mine starts by checking my phone. But where most people look for the weather update, local traffic, or even check Twitter or Facebook, I use my phone to peer an inch inside my daughter’s abdomen. There, a tiny electrochemical sensor continuously samples the fluid between her cells, measuring the concentration of glucose so that we can control the amount of insulin she’s receiving through her insulin pump.
Type 1 diabetes is a nasty disease, usually sprung on the victim early in life and making every day a series of medical procedures – calculating the correct amount of insulin to use for each morsel of food consumed, dealing with the inevitable high and low blood glucose readings, and pinprick after pinprick to test the blood. Continuous glucose monitoring (CGM) has been a godsend to us and millions of diabetic families, as it gives us the freedom to let our kids be kids and go on sleepovers and have one more slice of pizza without turning it into a major project. Plus, good control of blood glucose means less chance of the dire consequences of diabetes later in life, like blindness, heart disease, and amputations. And I have to say I think it’s pretty neat that I have telemetry on my child; we like to call her our “cyborg kid.”
But for all the benefits of CGM, it’s not without its downsides. It’s wickedly expensive in terms of consumables and electronics, it requires an invasive procedure to place sensors, and even in this age of tiny electronics, it’s still comparatively bulky. It seems like we should be a lot further along with the technology than we are, but as it turns out, CGM is actually pretty hard to do, and there are some pretty solid reasons why the technology seems stuck.
CGM systems are composed of three main pieces: the sensor itself, which changes glucose concentration in the body to an electrical signal; the transmitter, which conditions the signals and encodes it for wireless transmission; and the receiver, which can be a standalone unit or built right into an insulin pump, and displays the current readings and a chart of the glucose trend over the last 24 hours.
The sensor is the key to the whole thing. The chemistry behind it is simple: an ultrafine gold wire is covered with glucose oxidase, an enzyme derived from the bacteria Penicillium notatum. This enzyme oxidizes glucose into D-glucono-1,5-lactone and hydrogen peroxide. The peroxide then oxidizes on the gold wire, resulting in a current proportional to the glucose concentration in the interstitial fluid. This current is read by the system and used to calculate an estimate of the blood glucose concentration based on a calibration curve.
But while the chemistry is straightforward, human biology and manufacturing challenges make a practical CGM sensor difficult. First and foremost, CGM sensors must be inserted into the interstitial fluid and live there for up to a week (although many of us stretch that out considerably to save on sensors). As a foreign object inside the body, a wire coated with proteins derived from a bacterium would be red meat to the immune system, which is designed to mop up exactly this kind of foreign invader. Without some kind of protection, the glucose oxidase that makes the sensor work would be destroyed by the immune system in a matter of hours. This requires special, proprietary coatings over the sensor that can allow glucose in but prevent the immune system from attacking the enzyme, at least for a while.
The other difficulty involves handling tiny components and assembling them into an interface for the transmitter that will digitize the signals from the sensor and send them wirelessly to the receiver. The interface has to provide both a place for the transmitter to mate and a way to adhere to the skin reliably for weeks at a time without causing any kind of contact dermatitis or other side effects. The sensor also needs to be mated to some kind of introducer, a thicker hypodermic needle through which the fine, floppy sensor wire can pass, so it can be inserted without bending. The finished assembly also has to be sterilized, of course, so it has to be able to withstand the rigors of irradiation, the most common method of sterilizing medical devices.
All things considered, the $75 we pay for each CGM sensor probably isn’t too unreasonable. Where the value proposition starts to break down for me is the transmitters. Designs vary between manufacturers, and even within a manufacturer’s offerings as new technology supplants the older stuff. Schematics are hard to come by, of course, but the FCC ID database and a wealth of teardowns by frustrated users show that the basic guts of the transmitters we use are about what you’d expect from any wireless technology – signal conditioning for the sensor, a microcontroller, some power management stuff, and a wireless subsystem. The transmitters we’re using now use the 2.4-GHz ISM band; we’ve also used transmitters that speak Bluetooth, but those transmitters last only a quarter of the time the older ones do.
Looking at these teardowns, it’s hard to swallow that the $600 check I write to buy each one of these things is money well spent. We all know just how little these things actually cost to manufacture – probably far less than $5 a unit, and that’s really being generous. And yes, I know that I’m not paying for the components and the labor to put them together – I’m paying for the billions in R&D and clinical tests that it took to bring these devices to market. But I can’t help but think there has to be a better way.
The basic problem is that these transmitters are sealed units. I mean really sealed – the resin capsule is injection molded completely around the PCBs, with no way to open it non-destructively. And with no way to open the case, when the batteries inside finally die, all you can do is replace it. This hasn’t kept intrepid hackers from replacing the batteries, of course; I haven’t tried yet, but it’s on my to-do list. Once the sensor is opened up it’s pretty simple to slip in new batteries, but sealing the sensor again so that it stays waterproof is a bit of a challenge.
But does it have to be that way? Is it really necessary to have a relatively large, bulky transmitter like that? Seems to me that the transmitter could be much smaller and much cheaper if it used some sort of RFID technology. An external transmitter could interrogate the sensor and receive back the tiny bit of data needed to encode the current sensor voltage. That seemed to be the way Verily, the Google-owned medical devices company, was going with their sensor-equipped contact lenses for CGM. The idea seemed solid, since glucose is excreted in tears, and results in one of the early symptoms of Type 1 diabetes onset – blurry vision dues to sugar crystals on the cornea. The contacts would certainly not have a battery on board, so it would have to be remotely powered.
Sadly, though, Verily just announced they’re throwing in the towel on CGM contacts, stating that the biological hurdles to getting a stable, diagnostically useful reading were just too great to surmount. As we’ve seen, CGM is not easy, but I have a hard time believing that Verily bailed on this for purely technical reasons. With my cynical hat on, I’d say that the prospect of jumping through endless regulatory hoops was just too much for the company to bear, which is a crying shame (sorry) for such a seemingly breakthrough technology. But it’s tough to go up against established players that already have regulatory approvals and have deep pockets to boot.
Still, I’m hopeful that someone will pick up the challenge of building a better CGM. It’s not bad now, and at the end of the day I’d probably do whatever I had to do to keep my kid cyborged up. It’s just a shame that the CGM companies know this, and don’t feel particularly compelled to be a little more generous on pricing for a captive market.
Why is Continuous Glucose Monitoring So Hard?
Source: HackADay
