Diabetes patients’ DIY solutions are still the standard of care

Paging Dr. Internet, we need a diagnosis. In this series, Mashable examines the online world’s influence on our health and prescribes new ways forward.

For 33 years, Orla Wilson managed her Type 1 diabetes well. “I always had reasonable control,” she says. “I didn’t really think too much about it.”

But when her seven-year-old daughter Polly received the same diagnosis in 2014, she and her husband began losing sleep. They were taking turns waking up every three hours in the night to check Polly’s continuous glucose monitor (CGM), ensuring her blood sugar didn’t drop dangerously low in her sleep. Wilson, an associate professor of materials science and engineering at Johns Hopkins University, decided to search online diabetes forums for advice. That’s when she learned she could program a machine to take the night shift.

SEE ALSO: Best fitness trackers for monitoring heart rate

For Wilson and her daughter, living with Type 1 diabetes requires around-the-clock maintenance. Since their bodies don’t produce enough insulin, the hormone that fuels cells in the bloodstream with glucose, they need to inject it. Without the right amounts of insulin, glucose builds up in a diabetic’s bloodstream, unable to be absorbed by cells that need it to function, and their blood sugar can rise. Insulin is like the key that opens up the door, or membrane, of the cell. “Without that key, you can’t get the energy inside the cells, effectively starving them,” says Dr. Michael Haller, a pediatric endocrinology professor at the University of Florida Health.

Wilson, after hearing conversations about ongoing diabetes projects in the local Facebook group “Baltimore Parents of Kids with Type 1 Diabetes,” stumbled on a manual with step-by-step instructions to build a device that could administer her and her daughter’s insulin automatically—the Open Artificial Pancreas System, or OpenAPS. The project, created by patient inventor Dana Lewis and software engineers Scott Leibrand and Ben West in 2015, was the first open-source effort to create an artificial pancreas, coming together two years ahead of the first commercial system, Medtronic’s MiniMed 670G in 2017. In early 2016, Wilson began building her own OpenAPS, joining a growing number of DIYers who elect to build their own artificial pancreas, either because the diabetes device industry does not offer what they need or because what is offered is not affordable.

In order to keep their blood sugars at a safe level, “Type 1 patients have to do traditional doctor-level thinking processes multiple times a day,” Haller says, carefully weighing how everyday activities like eating, exercising, and sleeping will affect them. With more than 42 factors influencing glucose levels though, including stress, illness, even the time of day, “it’s nearly impossible to tease out all these factors,” says Dr. Laurel Messer, a clinical scientist at the Barbara Davis Center for Childhood Diabetes and the University of Colorado School of Medicine.

If they knocked at my front door, I wouldn’t know who they were, but they were always just so willing to help.

Starting in May 2016, Wilson entered hundreds of lines of code into the open-source software platform, Ubuntu, following the OpenAPS instruction manual. When she had to troubleshoot, she turned to an OpenAPS space on Gitter, a messaging platform for developers, to ask for “tech support.” It was a relief “how quickly responses would come from people all over the world. If they knocked at my front door, I wouldn’t know who they were,” she says, “but they were always just so willing to help.”

Late in October that year, she got the device working in her Baltimore home office. “I just sat there watching my computer screen,” she says, “absolutely amazed by what it was doing.” On her computer screen, lines of information were stacking up as the system she built went through “loops.” She had reverse-engineered an old Medtronic insulin pump through trial-and-error. Now it was hooked up to a small credit-card sized computer called a Raspberry Pi, in which she programmed an algorithm to tell the pump how much insulin to give her automatically. The Raspberry Pi, or “rig,” was downloading her blood glucose data every five minutes from her CGM, a patch she wears on her arm. She could see this data on her screen now, too. If the data showed that her glucose levels are rising, the rig told the insulin pump to dispense more insulin to her subcutaneously. If her glucose levels were going down, the rig told the insulin pump to preemptively dispense less insulin. This communication between the devices is called a “hybrid closed loop” or, in action, “looping.”

“For a long time, I never questioned why the CGM and insulin pump didn’t communicate with each other,” Wilson says. She had been using both for four years, checking her CGM nearly every hour in a day, doing the mental calculations for insulin, and then manually entering into the insulin pump what adjustments to make to her insulin levels. Now there was a rig to do this for her. “It’s running in the background all the time,” she says. “I have to tell it when I eat but it does everything else.”

Wilson’s first Open Artificial Pancreas System created in October 2016. Top to bottom: Dexcom receiver (displays blood glucose levels sent from the monitor on her arm), CareLink USB stick (communicates between the Raspberry Pi minicomputer and insulin pump) next to the Raspberry Pi, a battery, and Medtronic insulin pump.
Credit: Orla Wilson

After using the system safely for a few months, Wilson was able to convince her husband that it was safe enough for Polly to try one and she started on another OpenAPS. “Maybe once a month we have to do something in the middle of the night for her now,” Orla says. Once they were comfortable with Polly using it, they could sleep easy knowing it was watching her blood sugar .”It probably didn’t make a huge difference for me,” Wilson reflects,”but for our daughter, it made a very significant difference.”

A lot of assessments of closed loop systems leave out this variable of sleep, notes Joyce Lee, a pediatrics professor specializing in diabetes at the University of Michigan. She believes patients and caregivers using these systems are sleeping much better. Like Wilson and her husband, “they’re not doing corrections or dealing with insulin loads at night—the machine’s doing it,” Lee says. “The value is in the reduced amount of work and intervention.”

In fact, hybrid closed loop systems “allow more patients to achieve glycemic targets than anything else right now,” Haller says. Their glucose levels tend to be more stable, and they have fewer incidents of going too high or low. Current guidelines from the American Diabetes Associations suggest that individuals with Type 1 diabetes have an A1C, or average glucose levels, at 7 percent or less, where a lower percentage means a lower average. (For people living without diabetes, a normal A1C level is below 5.7 percent.) Time in range is another factor measuring how much time an individual spends within an ideal glucose range, usually 70 to 180 mg/dl.

My system is looking out for me.

Kyle Rose, a user of another DIY closed loop system called Loop for iPhone, in southern California, reported that once he began using his, he woke up for the first time in 25 years of having diabetes at his target of 100 mg/dl five days in a row. “My system is looking out for me,” he said.

A look at Orla Wilson’s Loop app, which she uses to control the amount of insulin that is delivered.
Credit: screenshot: Loop

According to Messer, adults using hybrid closed loops can achieve an A1C of about 7 percent, spending around 70 percent of their time in range.”This is very difficult to achieve without diabetes technologies,” she says. “To have the majority of individuals trying these devices be able to obtain glucose levels like that is quite a breakthrough. We’ve never had anything that effective in the past.”

The hybrid closed loop system is one innovation born out of a larger online movement, #WeAreNotWaiting, which appeared on Twitter first in 2013. The hashtag represents a deep undertaking by diabetes patients and caregivers with software expertise to overhaul diabetes technologies and make them widely accessible when device companies could not deliver solutions fast enough.

All these people in the diabetes community just took a piece of the problem and said, ‘I’m going to go figure that out.’

The group of DIYers and entrepreneurs “has driven companies to have better products much faster,” Haller says. Frustrated with the usability of device interfaces and with the industry’s proprietary software and data restrictions, they developed a host of platforms, apps, hardware, and cloud-based and solutions to improve their health outcomes.

Wilson’s Loop system that she and her daughter Polly have used for three years now. Left to right: Omnipod tubeless insulin pump, RileyLink (open-source hardware device) and battery contained in a 3D-printed case. The RileyLink bridges the differing radio frequencies of the Omnipod pump and her iPhone (not pictured) to communicate between them.
Credit: Orla wilson

“All these people in the diabetes community just took a piece of the problem and said, ‘I’m going to go figure that out,'” says Howard Look, founder and CEO of Tidepool, a nonprofit organization dedicated to making diabetes data more accessible. When his daughter was diagnosed with Type 1, he couldn’t view the data from her different devices on one platform and decided to write software to integrate those data streams. “The #WeAreNotWaiting movement is the leading edge of technical innovation,” he says. “They are looking ahead, asking what’s possible with the technology available to us, and doing the innovation themselves, releasing it freely and paying it forward.”

Wilson’s daughter Polly had a brief stint with a commercial system, a Tandem T:slim, but at the time, it could only respond to her blood sugar going low. They have since switched from OpenAPS to Loop, another DIY system that, instead of using a rig to control the insulin pump, uses an iPhone app. Look and his team at Tidepool are now in the process of receiving FDA approval for Loop to appear in the App Store. “If all our rigs fell apart, I would use [the Tandem],” Wilson says, “but for now, I’m quite happy with what I have.”

Jayne Williamson-Lee is a health and technology writer based in Denver. You can find her work on j-writer.com.