Diabetics have been waiting for years for better technology to manage their condition. Some got tired of waiting and hacked together an open source hardware and software solution. This is their story.
By Jo Best May 28, 2017
This is a story about what happens when people decide technology’s potential to help them is too great to leave in the hands of hardware companies. It’s about open source and commercial software, about the little guy and the regulator, about technology and your body.
For years, type 1 diabetics have been told that technology was the answer to their problems, that a solution could be found, that the daily grind of managing their condition could be fixed by the right mix of hardware and software. One day. One day soon. Just not yet.
“Diabetes sucks deeply, the technology we are given to manage it sucks deeply, and we are pretty much tired of waiting. We’ve been told a ‘cure’ (or at least, a mostly foolproof way to manage it) is just 5 years out. I’ve been told this, personally, every year for the last 25,” Scott Hanselman, a type 1 diabetic and technologist, wrote on his blog in June 2016.
But, after years of waiting for technology to do what people promised it would, something is starting to change.
“I’m actually feeling like we are on the edge of something big. I believe that now we are inside a five year window of time where we WILL make Type 1 Diabetes MUCH, MUCH easier to deal with,” Hanselman wrote.
So what’s changed? Diabetic technologists have stopped waiting for other people to make the tech they need, and started making it themselves.
The #WeAreNotWaiting movement
Artificial pancreas devices might hit the market in 2018. Maybe. But what if you’re one of the type 1 diabetics who really needs the technology now? Do you hold on, injecting yourself with insulin up to six times a day, or do you use the IT skills you already have to make a homemade version?
For the diabetics that are part of the “We Are Not Waiting” movement, there was no doubt in their minds that hacking together their own hardware and software to manage the chronic condition was the way forward.
The movement was born from frustration, particularly among the parents of young children with type 1 diabetes, at the pace at which technology that could potentially revolutionise their lives was being developed.
Diabetes may be a relatively common and well-known condition, but managing it on a day-to-day basis is anything but easy.
In type I diabetes, the body either doesn’t produce enough insulin—the chemical needed to keep blood sugar in the right range for health—or doesn’t respond to the insulin it does make. For diabetics, the condition means regularly monitoring their own blood sugar levels through a fingerstick test and then adjusting their insulin level, using frequent injections of the chemical, themselves.
For adults, it’s a tricky and time-consuming process. For children, it’s a whole other matter: their parents may have to wake them in the night—the time when their glucose level will typically fall—to monitor their glucose level and administer insulin accordingly, for example. It’s a routine that takes a toll on both the children and their parents.
One way type 1 diabetics have to control their blood sugar levels is through an insulin pump. Rather than having diabetics manually inject insulin, an insulin pump can deliver the hormone directly into the wearer’s body. It can also allow a more fine-grained approach to insulin delivery by varying the amount of the hormone and intervals between doses. That helps keep blood sugar more tightly within the desired range, but it can also mean pump users have to test their blood sugar more frequently that before.
A solution to that problem comes in the form of a continuous glucose monitor (CGM), which takes a reading of the user’s blood sugar levels every five minutes, giving them a near real-time view of how their glucose levels are changing, helping them refine how they use their pump.
However, with both the monitor and pump, users can find themselves adrift in information overload—too much data for them to do anything useful with it.
“The problem with the pump is it gives you so much control, it can become a bit too much. You get a pump and people say ‘Wow, you’re cured’, but my condition has just become ten times more complicated. If I invest time into it and learn how to use it and adapt it to customise it for myself, yes, it can help,” said Tim Omer, a type 1 diabetic, IT consultant, and health hacker. “I still have to manage it and work it.”
“A lot of people with a CGM get overwhelmed, and they end up getting overwhelmed, because they feel like there’s too much pressure on them. As a diabetic, you always feel like you’re being told off all the time—by healthcare, by your medical devices shouting at you. It’s great to have more information, but if you’re just getting yelled at all the time, it’s frustrating,” he added.
Another downside is that the kit is expensive to buy and run—and like many phones, some battery-powered components will come to the end of their life when their battery completes a certain number of charging cycles, despite the rest of the hardware remaining fully-functional. The battery can’t be replaced, so when it goes, the rest of the device goes with it.
One project borne of the We Are Not Waiting movement was Nightscout, an open source system originally built to help the parents of diabetic children get a better handle on their children’s condition by giving them remote access to the readings from the child’s CGM.
Nightscout allows data from the CGM to be published online, by connecting a phone with the Nightscout app installed to the receiver part of the CGM. Data from the CGM can then be viewed through the Nightscout website or any web-enabled device, be it another phone or a smartwatch. While intended to help parents monitor young children’s condition while they’re away from home, it’s also used by adult diabetics to get a more user-friendly display of their blood glucose data.
A similar DIY project, called xDrip, is a device that gathers the data from the sensor part of the Dexcom G4 glucose monitors, and transmits it via Bluetooth Low Energy to an Android app, or feeds through into the Nightscout system. The drip is made of four components that are soldered together at home, and cost around £40 in total. Together, they can fit inside a housing made of a Tic Tac box.
“That does two really important things,” says Omer, who built his own xDrip at home in a couple of hours. “It means I don’t have to buy the manufacturer’s receiver [for the CGM] so my costs are significantly reduced… And it picks up the signal and it relays it to my phone. I now have control of the data.”
Closing the loop
The We Are Not Waiting movement is also turning its attentions to closed-loop systems, where the glucose monitor and pump are able to communicate with each other to keep the wearer’s blood glucose more tightly inside the right range.
One such project is OpenAPS (APS stands for artificial pancreas system), co-founded by Dana Lewis, a type 1 diabetic who found the alarm on her CGM wasn’t loud enough to wake her if her blood glucose went dangerously low while she slept. She wanted the data from her device so she could hack together to something with an alarm loud enough that she wouldn’t sleep through it, but couldn’t extract the data from her device to make it.
After a fellow diabetic shared code with her that allowed her to get the real-time data off the CGM and build a louder alarm system using her phone and computer, she created a tool to alert friends and family if certain blood-glucose parameters were crossed.
From there, Lewis was able to build an algorithm that could make predictive recommendations about what would happen in the future based on the real-time data, and then by finding a tool that would allow that data to be communicated to the pump—creating the closed-loop artificial pancreas system OpenAPS.
OpenAPS uses data on the carbohydrates in the wearer’s food and their blood sugar level, runs it through the personalised algorithm that determines how much insulin they’ll need in the future to keep their blood glucose at the right level.
“The beauty of it is it provides a recommendation in real time with real-time data. While a person with diabetes may make that calculation a dozen times a day, the system is doing it every five minutes. If something starts happening, the system is able to react and give a recommendation a lot sooner than a human, who might not otherwise notice something is happening and take action,” Lewis said.
“It won’t sense something coming that you don’t know about, but it has that attention — it takes a reading every five minutes — unlike the person who might be in a meeting, or playing with their kids, or out running, and not necessarily wanting to think about diabetes all the time.”
OpenAPS systems can send the diabetic’s basal rate to the pump, which means it can only make small adjustments to their insulin dose—small adjustments that the user can undo if they want. The dose will act over half an hour, so if the pump fails for any reason, the pump will revert back to its standard operating procedure.
OpenAPS technologies are not only helping adults to get a better handle on their condition, some children are also using it, too.
“They spend a lot less time in the nurse’s office, and more time in the classroom. It’s helping them learn about how to treat their diabetes, they’re learning self-management skills much more quickly than they would have without the system,” Lewis said.
DANA LEWIS, CO-FOUNDER OF OPENAPS
As the parents can remotely see the data from their child’s system, they can speak to the teacher or the nurse if anything is going awry to head off problems. Similarly, data from both children and adults can be shared with clinicians to help give them a better idea about how an individual is managing their diabetes, and potentially refine basal rates for new diabetics.
The brave new world of healthcare hacking still has its problems. For one, it’s outside the traditional, regulated world of medicine. Commercial devices and software used to manage or treat medical conditions undergo lengthy clinical trials to assess their safety and benefits, and have to be regulated by the US Food and Drug Administration. That means they take far longer to reach the market, and are far more expensive. It also means they come with reasonable expectations of safety among users.
For those developing their own homemade devices, there’s none of that. People writing code and publishing advice on how to build systems are not regulated, but they can’t distribute their hardware or software without running afoul of the regulator. While they can publish their source code, offer tips and advice to others about how to build devices, but that’s about as far as they can take it.
That means anyone who wants such a device has to build it themselves at home—they have to be convinced about how safe it is, and be confident they have the necessary skills to do so.
They also need to be able to get their hands on the right hardware. The OpenAPS system needs certain older models of CGM that the manufacturer has stopped making. And you can see why, from the manufacturer’s point of view. After discovering a vulnerability that could allow the pump to be hijacked, they turned off the ability for the hardware to be commanded remotely for most models. Only some earlier versions still have that capability, and so can be used in OpenAPS systems.
So far, around 110 people have built their own OpenAPS systems, according to Lewis, and that number is growing. It has also inspired others to use it as the basis for new diabetic-aiding technologies. Omer used the algorithm to make an open loop artificial pancreas notification system based on an Android app he created called HAPP.
“I originally started the project as a mess-around, not thinking it would be useful, but once I had something quite crude up and working, it was incredibly useful. The system works every 15 minutes when it will crunch the data, and then my watch will vibrate and it will say ‘make this adjustment’. It was massively useful in the sense I didn’t have to sync and check and monitor stuff. I could let the system do that for me and just tell me when I need to action something,” Omer said. (Disclosure: Omer previously worked for CBS Interactive which owns TechRepublic and ZDNet).
For now, the FDA is taking a wait-and-see approach towards homemade diabetic tech, exercising what it calls “enforcement discretion”—keeping tabs on the healthcare hackers, monitoring the situation, and choosing not to take any action. That’s not to say the regulator doesn’t have its concerns about homemade tech.
What happens if the systems break down and users who haven’t had to inject insulin before aren’t confident enough to do so, for example. But, it hasn’t take any action against those that publish instructions on how to build devices, and has been actively engaging with the DIY community.
“We understand why people are doing it, but we want to make sure they do it safely,” Dr. Courtney Lias, director of the FDA’s Division of Chemistry and Toxicology Devices, told a conference recently.
The FDA is also working with those medical hardware companies that are seeking to bring artificial pancreas systems to market, going through the regulatory hoops that the DIYers don’t have to. Similarly, the regulator seems to be taking a pragmatic approach to getting commercial hardware out into the wild, tolerating a level of risk to ensure the systems can be launched.
“Artificial pancreas devices do not have to be perfect with zero risk to be beneficial,” Lias said. “The approval decision is a benefit/risk decision. We make this decision in the context of the high risks that people with diabetes face every day.”
Because of the inherent risks to the overall health of people with type 1 diabetes from their condition, and glucose that isn’t well-managed, the FDA is ready to accept systems that come with some degree of risk.
“The FDA is really supportive of that technology reaching people with type 1 diabetes… They’re saying the community is willing to accept things that aren’t bulletproof and we won’t stand in the way. It’s amazing,” Dr. Roman Havorka, who leads research into artificial pancreas systems at Cambridge University research, told ZDNet recently.
SCOTT HANSELMAN, MICROSOFT SOFTWARE ENGINEER
If all goes well, artificial pancreases built on the work of groups like Havorka’s could be approved by the FDA in 2017, with commercial units reaching the market the year after. That would mean that those who don’t feel they have the technology skills to build their own devices will have another option to get one, and healthcare providers will be able to fund those units getting into the hands of diabetics across the world.
Does that mean that an end to the We Are Not Waiting movement? Perhaps not. After all, diabetics in less developed parts of the world are still waiting for systems that they can afford, which the first versions of the commercial artificial pancreases won’t help. Others may want functionality that commercial systems don’t deliver, or to export the data in a way that manufacturers don’t allow.
Instead, it’s likely that, much like in the software world elsewhere, the majority of people prefer to get black-box systems from a single commercial provider that provides support and updates they need along the way. Others, however, will remain part of the open source community, putting together new systems and sharing them with others on the same path—despite the risks—because they’ve waited too long for the traditional channels to help them.