“A Roomba ate my pancreas!” It sounds like the plot of a weird sci-fi comedy. But in Dana Lewis’s life, this is just a normal day.
Lewis is one of the first people in America to create her own mechanical pancreas in an attempt to better manage her type 1 diabetes. (Her robotic vacuum cleaner keeps slurping up and choking on the
system’s many cables.)
Excellent piece about Dana Lewis and the do-it-yourself artificial pancreas system. The WNYC production is well done.
I find many facets of this development interesting. On the one hand the FDA is leery of citizen hackers fooling with so-called “class A medical devices.” One person in the piece is quoted as saying that the FDA is very concerned about adverse events and treating diabetes with insulin is one long series of adverse events. These home brewed systems are cutting down the number of adverse events. Diabetes is already dangerous. To do nothing and just wait for the manufacturers and regulators to all their ducks in a row, we have to live with it in the meantime.
I think this whole development of grassroots techno-advocacy is interesting to talk about. They produce a productive irritant to the mix. Bravo!
Actually this debate is at the heart of FDA’s entire philosophy. What get’s to decide about the tradeoff between benefits and the risks of adverse events. If I wish to accept more risk, why should the FDA decide for me? How does some “medical professional” know my values?
This is a particularly sore spot for me, because I’m very comfortable with technology. I would close the loop the day my CGM arrived in the mail.
I don’t think FDA is trying to regulate choice out of the equation. FDAs problem is that they’ve certified these devices and they are unsecured. If I can read my CGM, what’s to stop me from reading yours? If I can command my pump, what is stopping me from commanding yours?
I think FDA should be concerned. They messed up here and now they know it.
What I want is an open API. FDA seems more inclined to give the manufacturers leeway to implement Super Secret Proprietary Security (the technical term for this is stupid). This, to me, is the philosophical problem at the heart of FDA. Their eagerness to accommodate vendors is what got us to this situation in the first place and now they want to double down.
Grrr.
This debate is at the heart of so many arguments with the FDA. Another that’s near and dear to my heart, for example, is raw milk. In some states it’s completely illegal to sell it. In others, it’s legal only if the point-of-sale is the dairy farm. The farmer can’t legally deliver it, in a refrigerated truck, to a refrigerator at a centralized co-op drop-off point. It’s been deemed more safe for each individual person to go to the farm (possibly 20-30 miles) and pick up their individual gallon of milk and toss it in a cooler bag with an ice pack and drive home with it.
I’m sure it has less to do with our safety than it does the safety of the bottom line for Big Aggra. Just like this issue has probably got less to do with safety than the bottom line of Big Pharma.
But I’m very cynical.
Thanks for posting the remarkable piece about Dana Lewis, Scott Leibrand and the DYI AP system OpenAPS. The follow-up interview with Ben West is very interesting. Ben is credited with figuring out how to access older Medtronic pumps and for writing much of the code that enabled OpenAPS. His closing remarks are spot on:
“As a reminder, I’m doing this as a last resort. Why does the market create such an intense pressure that multiple patients are willing to hack their own devices? I don’t see any conflict with the FDA. Their mission is to promote and protect safety. At the core that’s what it’s all about for me. The reason I’m doing this is because I want better safety for myself. I’ve actually worked with the FDA twice, I’ve had two official meetings with them. I’m excited about the possibility of working with the FDA. But right now their process is optimized for the vendors and regulators. I think they need to include patients as equal stakeholders. In practice, that means providing a way for small DIY projects like ours to work together.”
As a side note, I’ve been running OpenAPS for 5 months now using a 5 year old pump, and a $35 computer. It is a DIY system, not easy to put together or maintain. But, it has remarkable safeguards and it reduces the risks associated with insulin delivery while improving the outcomes.
That eagerness has effects that extend far beyond technology, like implementing what are in effect protective tariffs without the need for legislation, in order to protect domestic manufacturers. Different rant for another time.
I love, love this homegrown AP movement but what happens when all the older hackable pump models conk out?
I am pretty sure that hackable older pumps will continue to exist for a while – at least until a viable AP is on the market. People have hacked not just Medtronic pumps but also Animas Ping and Omnipod versions. The difference is that the safety features aren’t as robust and that the coding is much more extensive and not documented. So that’s much more DIY than the openAPS reference design.
Do you know if anyone has shared instructions for hacking Omnipod?
Last I saw, they were still working on it. Someone has actually put out a bounty on cracking the thing.
http://www.openomni.org/
I hope first gen commercial AP’s don’t set targets too high due to FDA “safety” concerns. The lack of common sense features and the ergonomic fails in pump design make me hope diy AP will remain a viable option for years to come. Ironically, the ability of AP to adjust insulin on the fly will obviate the need for what should already be standard on pumps such as Super Bolusing/Super Correcting.
Thanks for the link. I hope someone takes that bounty soon!
Well done Podcast on the topic! Thanks for sharing. None of it was exactly news to me, but I really enjoyed how the story was told and the information was presented.
It’s seems pretty amazing. Do you mind a few questions; Are you able to bolus from it or does it do basal adjustments only? How portable is it? Are you sporting a bag with battery, RPi, etc. or on a belt? How are your blood sugars since going on it? Is it dependant on nightscout? Thanks and hope it’s working wonderfully for you.
Good questions! Here is my Nightscout screenshot from last night, and a photo of the OpenAPS rig:
My OpenAPS rig is still the original version: Dexcom G4, Medtronic CareLink USB stick, and a Raspberry PI computer. Many people have built much smaller versions, such as the one shown in this recent WSJ article. I’ll be working on a more compact setup this summer. Every 5 minutes, the APS pulls pump history to compute IOB and, based on CGM data, calculates a correction in the form of a temp basal. For safety, the APS never issues boluses, and a temp basal is set to expire in 30 min. So, if anything fails, the system defaults back to standard pump operation. Also, it does not have to run all the time - I can turn it on or off any time I like. One still has to bolus for meals, but recent algorithm updates include some nifty meal assist features. The operation does not depend on Nightscout, but Nightscout provides a really nice way to visualize what’s going on. In the example above, I had OpenAPS working from around 9:30pm to around 6:30am, and its temp basal rates are shown in blue. During that time my lowest BG was 75 and the highest was 104. Pretty much every single night looks like that. My overall time in range is around 95% with less than 1% below 60 (for reference: I eat carbs in moderation (around 100g per day), and I do not think I have endogenous insulin production of any significance). Overall, OpenAPS has been an incredibly fun project for me!
Thanks for the post! This is incredible. One question–do you think it would be G5 compatible? Or do you need G4?
I’ve not upgraded to G5 yet, so I have not looked into that very carefully. My understanding is that G5 has been hacked (relatively recently), and that several people already have versions of OpenAPS running with G5.
Yes, it’s most certainly G5 compatible. That’s what I’m using.
Hi Melissa, are you connected through the receiver or from the phone?
