June 28, 2016 -- At least 85 people and counting are managing their type 1 diabetes with an artificial pancreas system they built themselves.
Dana Lewis is one of them.
The Seattle-based data analyst created a system with her husband Scott Leibrand, a computer network and systems engineer, in 2014.
Since then, she rarely has emergencies where she has to eat carbs because her blood sugar drops a lot. With that peace of mind, she sleeps better at night.
The system “has completely changed my life with diabetes,” Lewis says. “The obvious outcome is improved blood glucose, but the reason I originally started it is I wanted the freedom and the safety to fall asleep every night and not worry about dying in my sleep.”
Lewis belongs to a growing number of people hacking into and rigging up their monitoring and insulin delivery devices, called the “We Are Not Waiting” movement. Rather than wait for the FDA to approve available technology that can improve the quality of their lives, the scrappy group of movers and shakers is coming up with its own innovative solutions for managing various chronic conditions.
“The We Are Not Waiting movement is patients saying, you [FDA and manufacturers] have to make this possible, but if you’re not, we’re not going to wait,” Lewis says.
And now, they have some research to prove that their devices work.
Eighteen people who built their own systems were included in a study run by Lewis and Leibrand that they presented earlier this month at the American Diabetes Association’s annual meeting held in New Orleans.
The first-of-its-kind study showed that for people using the artificial pancreas, their blood sugar was more often within the goal range, their A1c (average blood sugar over a period of months) was lower, and most participants slept better. All had previously used an insulin pump and monitor to control their blood sugar levels.
Although promising, more research needs to be done with larger groups of people to know if they truly perform better.
“Users caution that DIY APS implementations require significant effort to build and maintain, and pointed out that these systems cannot be considered a 'technological cure,' but were extremely satisfied with the 'life changing' improvements associated with using an APS,” Lewis and Lebrand wrote in the study.
The wheels seem to be in motion for approval of a commercially available artificial pancreas system. The FDA recently released a draft guidance about creating glucose monitors and pumps that can talk to one another. As many as 10 artificial pancreas devices intended for commercial use are in the works. The first could hit the market in the spring of 2017. Researchers and manufacturers have been working out the algorithms for 10 to 15 years, according to the FDA.
Here’s what makes the artificial pancreas a game-changer.
A healthy pancreas makes insulin, a hormone that lowers sugar levels in blood. It gets sugar to cells in your body to be used as energy.
The pancreas doesn’t make insulin in people who have type 1 diabetes. They have to check their blood sugar throughout the day to figure out how much insulin they need at mealtimes, at bedtime, and other times that their blood sugar might fluctuate. They are doing math in their heads all day, and their lives depend on getting it right.
Some people inject insulin. Others use an insulin pump that delivers the hormone through a tiny needle left in place at all times, unless the user removes it. This saves the user from repeated daily shots.
Either way, the person with diabetes is doing the work of the pancreas by deciding how much insulin they need and doling it out.
For people using a continuous glucose monitor -- a device that provides a blood sugar reading every 5 minutes through a tiny sensor inserted under the skin -- an alarm sounds if their blood sugar begins to fall. But Lewis has slept through her alarm several times. And there was no way to raise the volume.
In 2013, Lewis decided she wanted to find a way to send readings from her monitor to her smartphone so that her phone would ring when her blood sugar was out of range. But her device didn’t allow users to export the data.
She and her now-husband Leibrand got programming code from John Costik, a dad who had hacked into his son’s glucose monitor so that he could track his son’s blood sugar on his smartwatch while he was at school. (Today, this technology is readily available without hacking.) Lewis and Leibrand used the code as a starting point to customize Lewis’s system.
At first, they used the data to send alerts to Lewis’s iPad and her husband’s smart phone if her numbers went too high or low.
Next, they created algorithms that would help predict blood sugar highs and lows before they happened.
But the holy grail was to figure out how to send these recommendations directly to the insulin pump. The couple found someone else with type 1 diabetes, Ben West in California, who shared the code to send commands directly to the insulin pump. The couple used the code to create a rig that could go wherever Lewis goes.
Their setup includes a continuous glucose monitor, (about $1000 to $1400 plus $35 to $100 per week for supplies); an insulin pump (about $5500 plus $100 per month for supplies); a pocket-sized computer called a Raspberry Pi (about $35 on Amazon); and a Carelink USB stick (about $100 on Amazon).
Now, Lewis’s glucose monitor sends her readings to the Raspberry Pi, which uses Lewis's and Leibrand’s algorithm to calculate how much insulin she needs and sends the command automatically to her insulin pump. Information is sent to and from the insulin pump through the Carelink USB stick, a device used to send data from standalone insulin pumps to a computer.
“Whether your blood sugar’s dropping, it’s flat, or it’s high, the system is constantly making those adjustments. That constant fine tuning simulates what a real pancreas does in the body,” Lewis says. Now that Lewis’s devices talk to each other, simulating the work of an actual pancreas, she rarely needs to “manually” correct her blood sugar by eating carbs.
“The number of times I have to take carbs for a low during the day has been reduced significantly,” Lewis says. “And I can’t remember a time in the last 6 months that I’ve woken up because of low blood sugar and had to take in carbs. It’s maybe happened once, whereas it would have happened hundreds of times before having [the artificial pancreas].”
She wears the pump and monitor, and keeps the rest of the system nearby -- on a desk or in a backpack, for example. If the system ever fails, the monitor and insulin pump work separately as they did before.
Lewis has been using the system for about a year and a half now. She, Leibrand, West, and others who have contributed code and technical solutions to the endeavor make the documentation publicly available online for anyone who wants to create their own artificial pancreas. That openness has helped 84 other people with type 1 diabetes from all over the world rig their own devices in the last year and a half.
But the do-it-yourself artificial pancreas may not be for everyone. Lewis stresses the DIY nature of the device when anyone asks her if they should build one, too. She tells them they don’t need to be programmers or engineers, but they need to be willing to learn many new things. Doctors’ initial excitement about the technology is sometimes tempered by the devices’ necessary tech savvy.
“These devices are incredibly complicated,” says Anne Peters, MD, director of University of Southern California clinical diabetes program. “As a provider, I’m incredibly excited about this because of what I think it can do.”
Peters says she would recommend them first for someone whose blood sugar levels aren’t well-controlled. And, she adds, “you’ve got to be able to use and understand the technology and troubleshoot all the devices.”
Lewis makes clear that she can’t build an artificial pancreas system for anyone else. “As a patient,” she writes on her blog, “I can only design tools and technology for myself. Because it would be seen by the FDA as a medical device, I cannot distribute it to other people to use, as it would have to first be reviewed and regulated by the FDA.”
The FDA emphasized this point in a webinar for the diabetes community earlier this month. Courtney Lias, director of the Division of Chemistry and Toxicology Devices within the Office of In Vitro Diagnostics and Radiological Devices, says it’s OK in the eyes of the FDA to develop your own device.
But “as soon as we get to the point where somebody is distributing information about automated insulin dosing or making artificial pancreas systems [for distribution], that’s when we feel the risk is higher and we may enforce,” Lias says.
So Lewis has only built an artificial pancreas for herself. But that hasn’t stopped her and others from publishing their code online for others to do with as they wish. Because she is not selling the information or distributing medical devices, the FDA cannot regulate her. The code she and others have published online is considered free speech. Lewis and Leibrand never aimed to patent a device and get rich off of it.
“We felt that going open source would help fill the gap between now and a good option coming to the commercial market,” Lewis says.