Dana Lewis at #HIC18

I was ecstatic to learn that Dana Lewis, who designed the OpenAPS system with Scott Leibrand, was going to be a keynote speaker at the Australian Health Informatics conference #HIC18 in Sydney from July 29-August 1.

It was fantastic to meet her and Scott and the brilliant group of Sydney closed loopers at the Opera House bar on Saturday 28 and on Tuesday I was able to sit in on her talk.

Patient as innovator

It was a breath of fresh air to hear Dana Lewis talk about patients as experts and innovators.

IMG_6960
Dana’s t-shirt has the OpenAPS oref0-determine-basal code on it

In her own words:

“We get a lot of pushback. We are told ‘you are not a clinician, you are not an IT expert, you are not an engineer. You are not a this or a that’. We often focus on the roles. Do your role, don’t cross the lines.”

IMG_6944

“And as a patient that’s really frustrating because as a patient we get that we are not professionally working in this space being paid to do this but this doesn’t mean that we’re not engaging and designing and developing and doing research and science.

We have to. This is making our lives better. Our other choice is to sit there and do nothing and suffer and wait for one day when something will be better.”

Patient as expert

IMG_6950
What we think everyone else knows – What people think ‘professionals’ know – What patients actually know

“The patient syndrome trap is hugely prevalent everywhere around the world in healthcare. We assume patients know little tiny bits of information and healthcare providers and everyone else knows way more. But that’s wrong.

The reality is that yes health care providers, IT, medical device companies, pharma and the government know a lot but we patients know so much more.

The difference is often in the process, the priorities.

Instead of focusing on the overlap and talking about whose role is doing what, the best partnerships, the best healthcare is going to happen, we get the best of all worlds when we all work together.”

Learning not to wait

Dana is currently principal investigator on a research project “Learning to not wait: Opening pathways for discovery, research, and innovation in health and healthcare.” She is researching how the data and innovation approach the #WeAreNotWaiting community has taken in diabetes can help other patient communities

You can see her posters and presentations from the 2018 American Diabetes Association Scientific Sessions at her DIYPS blog:

Other patients who are not waiting

Mike Morris has written a brilliant post on the Health Intersections blog about how he got cancer and decided he also was not waiting. He created his own integrated digital health dashboard. This meant his data held in various types of health records could be consolidated and available in near real-time so that he could work with his oncologist to course-correct in the middle of chemo cycles rather than going through whole chemo cycles and then waiting for a scan. He talks about having ‘personalised medicine’ rather than ‘a life of chemo’ and about treatments based on predictive outcomes and trends. Sound familiar?

His story and his efforts to create a better system remind me so much of Nightscout and the open source artificial pancreas community #WeAreNotWaiting. His diagram of the protocol treatment chemo cycle showing the 10 minute doctor’s visits followed by the expanse of time between visits where there is no monitoring reminded me a lot of traditional type one diabetes healthcare. It is very exciting to see this changing.

Tidepool: step right up!

Very interesting news that Tidepool, an open data not-for-profit made up of people who either have type one diabetes or are related to someone who has type one diabetes, is to deliver an officially approved, FDA regulated app based on the DIY tech of Loop. The fact that  Katie DiSimone and Pete Schwamb have just joined the Tidepool team is fascinating too. Patients as experts and innovators? Absolutely.

And what about my rant in my initial post about interoperability?

As Howard Look of Tidepool says “this pushes everyone forward, to thinking about interoperability and interchangeability, and that makes for a better world for people with diabetes.” What a sigh of relief after all these years. It’s going to be very interesting to watch the diabetes tech space over the coming years.

 

Consumer co-design

There was a lot of talk about co-designing diabetes health systems with patients at #HIC18 and I was pleased to see a session on consumer co-design scheduled at the Australian Diabetes Congress (August 22-24) in Adelaide.

Renza
Renza Scibilia T1 consumer voice at Diabetes Australia was one of the chairs in the Australian workshop on co-design

I was not able to attend the conference but I’m sure the main messages were summed up by this slide:

ConsumerCoDesignAtADC

User experience design for medical devices

I was even more thrilled a few months ago to see BigfootBiomedical advertising for user experience professionals to join its company. Doing proper ethnographic/observational studies of people actually using medical devices is imperative and I get the feeling it has been overlooked by some companies in the past.

With more vendors coming into the market with their versions of hybrid closed loop technology for diabetes in the imminent future, having good usability and the least possible diabetes burden may well determine which companies survive and thrive in the new marketplace.

It will also be interesting to see how companies that choose to embrace JDRF’s call for open protocols fare in the marketplace too. Interoperability, open source, community… perhaps a whole new health ecosystem? Who knows?

One thing for sure is, from my perspective, it’s going to be fascinating.

Now, back to the ADC co-design workshop. Other presenters in this workshop include:

KirstineBell
Kirstine Bell, dietician, diabetes educator and researcher
FrankSita
Frank Sita T1D blogger at Type1Writes

 

MelindaSeed
Melinda Seed T1D blogger and advocate at twicediabetes.com

… and diabetes educator/pharmacist Theresa Di Franco from Perth Diabetes Care

DIY hybrid closed looping

Another brilliant presentation at the Australian Diabetes Congress was Renza Scibilia, Cheryl Steele and David Burren’s talk on using DIY hybrid closed loop systems.

Renza uses Loop and talks about what this tech has meant to her, David uses Android APS and gives an overview of the technology and Cheryl uses OpenAPS and shows how Nightscout reports can be used by health care professionals. At the end of the presentation, CEO of Diabetes Australia, Greg Johnson, launches the Diabetes Australia position statement on DIY looping.

You can see a recording of the presentation here on Youtube.

CherylSteele
Cheryl Steele T1D diabetes educator extraordinaire and the person who got me started on insulin pumps
DavidBurren
David Burren T1D tech expert and photographer who blogs at BionicWookie

Next … Dana Lewis at #HIC18

The verdict is in!

Drumroll please…

Results

Three months prior to looping › first three months of looping

  • Predicted HbA1C 8.2 › 6.8
  • Time in range of 43% › 70% (76% in Nightscout)
  • Average glucose 10.5 › 8.2 (8.3 in Nightscout)
  • Standard deviation 3.6 › 2.6
  • Lows 1% › 1%

Before

Dexcom Clarity report

ThreeMonthsPreLooping

After

Nightscout report

20180816NightscoutCGMresults

So far so good

For many people a HbA1C of 6.8 would still be troublingly high. For me it is a major breakthrough, especially with my time in range increasing from 43% to 76%.

As Neil McLagan, a T1D athlete who’s not looping but has found stability on a low carb diet, said to me, “Before it was like, I was always going through the zone. I was high going through the zone on my way down or low going through the zone on the way up. Now, I’m always in the zone and it’s a good feeling.”

For me OpenAPS is a self-correcting system in that it gives me enough of a feeling of control (not just BG control) that I have the confidence, the breathing space, and hopefully the tools, to  make further changes to keep my blood sugars in range even more of the time by adjusting my settings and modifying some of my diabetes behaviours.

It’s like another new looper on the Facebook Looped group said, “I’m ten fold more motivated now.”

Due to the instability of my blood sugar levels prior to looping and my high insulin sensitivity, a lot of my diabetes management over the years has involved hypo-avoidance strategies. Having a system that helps me head off hypos by adjusting insulin delivery to match BG data every five minutes, and a CGM that gives accurate warning of hypos means I can surf closer to the ideal blood sugar range without freaking out.

IMG_8511
Me with my lovely man. My OpenAPS rig is in the little swing bag.

It is very motivating seeing results, after so many years of putting in effort for little return.

In fact five days after posting the three month results above I checked my reports again and found my predicted HbA1C had decreased to 6.3 with an average BG of 7.5, standard deviation 2.2 and 84% of time in range over the past two weeks. So it’s very encouraging.

LastTwoWeeksResults

Compare that with this. The same record for the three  months before I starting using OpenAPS

PreLoopingAGPDexcom

Thank you to everyone who offers their support and knowledge for this DIY technology on social media, blogs and gitter. It has been mind-blowing to find such a savvy and supportive community after all these years.

And once again, I will always be grateful to the pioneers who said #WeAreNotWaiting and found a way to make it happen. The sheer dedication and years of hard work involved are staggering to contemplate.

I’m planning to keep this site updated with the strategies I’ve tried and the things that have worked for me. Everyone’s diabetes is different, this is no walk in the park, and I realise I have been lucky to have a very supportive and technically knowledgeable partner, but by sharing my experiences I hope others can learn from them too.

Hold on a minute… it gets better... One week later …

10 September 2018

The last week’s predicted HbA1C is:

  • 6.0 with 1.2% of CGM readings less than 4mmol/L, 94.2% time in range and down to 1.6% standard deviation.

1.2 percent hypo_2018-09-10 at 6.40.01 pm

  • Or if we change the hypo range to being less than 3.9, we get 0.8% readings less than 3.9

0.8Hypos2018-09-10 at 6.41.15 pm.png

  • 0.1% of readings less than 3.0mmol/L with 95.3% time in range

… Very happy camper… you can see why this is motivating!

Next … Consumer co-design 

 

Starting to loop

A frustrating start

I was finally on the verge of looping and yet the last bit of the wait seemed the longest. We had all the pieces and my partner Michael had installed everything. But it wouldn’t work. It took us awhile to figure out that the Edison I had required an external antenna! One more thing to order and wait for.

6 May 2018

It arrived. We hooked everything up and switched it on! Woo Hoo! Computer code appears, huge unstoppable grins appear. We have lift off.

IMG_4247

It was amazing to watch the algorithm do its stuff. Supporting me and my blood sugars. Making cool, calm mathematical predictions and adjusting my insulin pump accordingly. Finally it wasn’t just up to me.

And if something went wrong it wasn’t my fault! Just some settings that needed tweaking. Some code that needed running or reinstalling. Some wires that needed connecting. It was now all about the technology supporting a spontaneous life, not a life fitting in around the technology.

Ok that is a bit of an overstatement. The system does require a lot of vigilance, user behaviour modification and sometimes relentless trouble-shooting. But it did shift the focus. I loved it!

I was so excited I started texting my old diabetes pals, torturing them with photos of the brain itself and my blood sugar prediction graphs.

I did a happy dance for the next three months non-stop. For the first time in 38 years it felt like something exciting had happened in my diabetes. With that came a much-needed surge of hope, joy, confidence and well-being.

IMG_6997

There were lots of connectivity issues over the first few weeks and a major learning curve that I hadn’t anticipated. A new update to the OpenAPS algorithm sorted out the connectivity problems I was having at home but I still had issues at work.

When I was at work (in a high school for the first three months of looping) I depended on a mobile signal to keep my system running.  Unfortunately in many of the classrooms I worked in the signal was too weak and the loop dropped out. There are offline looping solutions for OpenAPS but I wasn’t able to get them up and running at that time.

Bolusing for meals

Another thing to adjust to was no longer being able to use square wave or dual wave boluses. I’d been having a lot of success with a 20% 80% dual wave bolus over a one hour period for most meals before I started looping. I could still use this method but it meant that OpenAPS couldn’t run and control my blood sugar levels during that hour.

IMG_7004
Quinoa, lentils, potato.. yum… you tell me how many carbs are in that meal? The joy of the carbohydrate lucky dip strikes again!

I experimented with various strategies and started keeping a food diary in my phone to record the outcome of using various bolus techniques with different food. This is still a work in progress.

My general approach is to give a normal bolus for meals containing fairly fast acting carbs (eg jasmine rice) and for other meals to give between 20% and 75% of the bolus insulin (depending on the meal) ‘up front’ or just as I start eating and allow the remainder of the carbs to be covered by the OpenAPS algorithm.

This is still hit and miss for me at the moment. Sometimes I’ve found it helpful to use the ‘Eating Soon’ mode of OpenAPS before meals. It sets a lower temporary BG target and means there is some insulin on board when the carbs hit to start working with them and it can help prevent post-meal spikes. But sometimes this mode brings me too low. Trial and error. YDMV (Your diabetes may vary).

Fiasp, where are you?

I am looking forward to Fiasp, super fast acting insulin, becoming available in Australia as many people report a reduction in post-meal spikes with it. One teenage boy using Fiasp in the US, apparently often doesn’t bolus for meals and still gets a HbA1C of 6.2 using the OpenAPS system.

Fiasp insulin

To me that is the Holy Grail of this technology. When a teenager with diabetes can eat normally… ie is ravenously hungry during a growth spurt and can eat three bowls of cereal followed by pancakes with maple syrup… and still have decent blood sugars.

Getting through those years without diabetes damage to the body and without the assault to the psyche years later when complications hit, is the major triumph of this hybrid closed-loop technology.

Treating lows

With my BGs now in range so much more of the time I found I only needed small amounts of fast-acting carbohydrate to bump my BGs up when they dropped below 4.0. I now have one or two of the 4g glucose tablets or half a Hypopak.

fullsizeoutput_4f49.jpeg

Low carb

I’ve been inspired by what I’ve learned lately about low carb diets for blood sugar control and am currently trying to have one or two meals a day with very low carb content. It makes so much sense.

But there will always be a place in my life for delicious bowls of steaming pesto linguine or sage and pumpkin gnocchi. Absolutely thrilled to say that OpenAPS handles these slow absorbing carb meals beautifully.

Next … The verdict is in!

 

 

Refresher course in T1D

As well as logging my data in Nightscout and trying to be more mindful of everything I did with my diabetes, I thought it would be worth a visit or three to a diabetes educator.

But it had to be the right one…

I had seen an absolutely lovely woman at my diabetes clinic over a decade ago. I was the first person to go on a pump there in 2000 and she helped me upgrade that first pump in 2004. She was in private practice now and I decided to track her down again.

IMG_5996
Here she is showing me the correct way to take the reservoir out of the connector

CGM was the key

I loved my endocrinologist and had been seeing him for 18 years. He had supported me through two pregnancies, was always available to answer my questions via email and was an advocate for people with diabetes being more knowledgeable about their own diabetes than their care providers. He encouraged me to read research articles, understood my fear of hypos and tried to tread gently with his suggestions. But still, even with a fantastic endocrinologist in a publicly-funded and accessible health system, I struggled to achieve healthy blood sugar levels.

It was like a dance. I’d go and see him every four months and he’d download my pump and BG meter data, look for patterns and advise me to increase my basal and bolus rates by just the smallest amount. I’d do that and leave the clinic, often start having hypos, and then switch the rates back to what I’d used previously.

What I now realise is that two things were actually changing after those endocrinology appointments, my settings which had been adjusted, and my behaviour. For the few days after the clinic appointment I would force myself to use the bolus wizard (which I did not usually trust), I would try to be more accurate with my carb counting and would try to bolus for meals earlier. So it was the double whammy of adjusted settings and behaviour change that often led to the hypos, not the settings adjustments alone. Also, because I was not using the pump’s Bolus Wizard during the four months between appointments, my endocrinologist did not have accurate information on what I was doing in terms of meal bolusing and carbohydrates.

My data always looked like a complete scramble to me and I found it extremely difficult to see patterns in it. I had a lot of variability from day to day.

This was before CGM.

The diabetes educator made an interesting comment to me on that first visit. She had downloaded my pump data (which included BG readings from my linked meter) and she also downloaded my CGM data from a separate system.

She said the advice she gave me based on the CGM data was very different to the advice she would have given me if she’d only had the pump/BG meter data. So there it was. At least part of the explanation for why my years of visiting my diabetes clinic had been less than optimal in terms of sorting out my BGs. No CGM data.

I was also incredibly sensitive to insulin and it was the hypo avoiding or minimising capabilities of the hybrid closed loop system I was about to try that would make the real difference.

Fix the basic stuff first

I had picked up some habits over the years in relation to my T1D management that weren’t working for me and I didn’t even realise it.

The first one was not realising that many of my pump sites were dodgy.

Bent cannulas

I’d been using the Medtronic Quicksets since I started pumping and inserting using a Quick-serter. My Quick-serter was ancient! These work brilliantly for most people but I didn’t have a lot of fat and kinked cannulas were a regular occurrence.

IMG_5792.JPG

First I bought a new Medtronic inserter which seemed more gentle. But then I followed the diabetes educator’s advice and tried other types of sets. I tried everything on offer from Medtronic.

I found the best sets for me were the 6mm Mios. I still get dodgy sites from time to time where there is some problem with absorption. But I’ve only had one bent cannula in six months.

PinkMio

David Burren has a great dual cannula hack for not getting stuck with bent cannulas. It basically involves leaving your old site in place for a few hours after inserting and connecting to the new one. By then you will know if your BGs are skyrocketing due to a bent cannula or dodgy site and it means you can reconnect to the old site to get the insulin flowing immediately and sort out a set change when it’s next convenient.

Rotate your sites

I guess this is a no brainer but I wasn’t doing it. I’d gotten into the habit of using my buttocks area during my first pregnancy 14 years ago and had stuck with it. It led to less bent cannulas than the abdomen. The diabetes educator implored me to start using my abdomen to give the other area a break. It made a big difference to BG control.

I think in hindsight I probably had a lot of variable insulin absorption due to overuse of the same sites. It’s probably one of the major issues for those of us who’ve been pumping with T1D for a long time. Preserve that site real estate at all costs!

TummySites
I try to do something like this. Image from http://www.clinidiabet.com

The bottom line is, now that I’m looping with OpenAPS I detect insulin absorption problems much faster than I did pre-looping. But having accurate CGM readings and good functional pump sites is crucial to the system working optimally. I found it was a great time to sort out this stuff while I was waiting for my looping gear to arrive.

Next … Starting to loop

 

Getting ready to loop

I ordered the parts for my OpenAPS rig and waited impatiently for them to arrive. I couldn’t wait to get started!

Piece by piece, the life-changing system arrived at my front door. In unassuming brown boxes from around the world.

IMG_4150

I didn’t have a compatible insulin pump despite pumping for 18 years so the hunt for a pump began. This was pretty gruelling but eventually after three months of searching I found a Medtronic 522 on an online forum. To say the hunt for the pump was an emotional rollercoaster is an understatement.

Disclaimer: OpenAPS hybrid closed looping is not for the feint-hearted. I am immensely grateful to my partner Michael for setting the system up for me. It was way more labour-intensive than I’d imagined. He has a PhD in computer science and is one of the most tech savvy people I know. But still, we struggled with aspects of the setup and it’s been a gradual process of trouble shooting, fixes and tension at times. But well worth it.

Starting CGM

There is no government subsidy or health insurance rebate available for adults using continuous glucose monitoring systems in Australia at the time of writing of this post. The costs are prohibitive for most people. I hope this situation changes because the potential health improvements and long-term cost savings for the health system overall are enormous. I would love everyone to be able to benefit from this technology.

MyFirstDexcomG5Sensor
My first Dexcom G5 sensor. This is a $92 piece of equipment – I don’t want to stuff it up

I had trialled Dexcom’s G4 system in 2013 and found it incredibly accurate. Unfortunately I could not afford to use it at that time.

I started using Dexcom’s G5 system continuously in March 2018 and was relieved to find the system was as accurate as ever.

The first thing I noticed was my morning blood glucose spikes. It didn’t matter whether I woke at 4am, 6am, 8am… It was nothing to do with food or coffee, my basal settings were not wrong. This was a 4-6mmol increase in blood glucose levels caused by a surge of ‘getting up’ hormones. Dana Lewis has written about the same phenomenon in her DIY pancreas blog.

The other undeniable thing I noticed looking at my CGM readings was the carbohydrate rollercoaster. It was very confronting to see just how wild those swings were.

Nightscout

The next step for me was using Nightscout. A fantastic cloud-based platform set up by parents of children with diabetes to visualise and monitor CGM data that also allows entries for all sorts of diabetes treatments. I loved all the detail I could capture with this system. I’d been wanting to interact with my data like this for years.

Nightscout Screen Shot 2018-08-12 at 6.51.33 am
Notice the morning spike

Close up of log a treatment 2018-08-12 at 6.57.14 am

For two months before I set up OpenAPS I did my best to track my food, exercise and bolus insulin through the Nightscout website. It was very challenging because I had to enter everything manually at this point. Every night I looked through my pump bolus history and added the data to Nightscout.

I did this for a few reasons.

  1. To capture data about what my blood glucose levels were doing pre-looping.
  2. To get data for fine-tuning my basal rates, carb:insulin ratio and insulin sensitivity factor (ISF was what I’d previously called my correction factor).
  3. So that I could start using Autotune a system developed by Dana Lewis and Scott Leibrand which iteratively calculates adjustments to basal, ISF and C:I ratios based on observed data.
Autotune2018-08-12 at 7.23.25 am
Today’s  autotune suggestions

Next… Refresher course in T1D

The dawn of the artificial pancreas

Within weeks of writing that first post in 2017 I realised everything had changed. Medtronic had launched their 670G hybrid closed loop system in the US and pivotal trials were underway for systems by Bigfoot Biomedical, Beta Bionics, Insulet, Diabeloop and TypeZero/Tandem/Dexcom.

The dawn of the artificial pancreas had well and truly begun. It was as though I’d been underground for the past fifteen years and hadn’t realised a revolution was happening around me.

#WeAreNotWaiting

Nightscoutlogo

It turned out that a group of super-smart and dedicated DIY engineers and programmers had been working nights and weekends for years to reverse-engineer old insulin pump protocols and create open source software that anyone could use to get their CGMs and pumps to talk to each other via the cloud.

These were people with type one diabetes themselves or people who had kids with type one. People who got it.

#OpenAPS

Thanks to my tech savvy partner Michael I was able to start using OpenAPS in May 2018. My predicted HbA1C based on Dexcom readings went from 8.2 to 6.8 in the first six weeks of using the system. My time in range went from 43% to 70%. Hypos were minimal.

My 82 year old mother had tears in her eyes when I told her on Mothers Day that I had an artificial pancreas. It had been a long wait for her too!

I will always be grateful to the pioneers of this technology and the quiet army of people testing, advancing and supporting this incredible DIY community. They have made their technology available to everyone who wants and is able to use it, including medical technology companies. There is no doubt in my mind that their efforts have ramped up the speed up of development of commercial products as well as fast-tracked the approval of new products through the FDA.

For me it’s been nothing short of a miracle.

This blog is an attempt to capture what’s led me to this point. Some of the steps I’ve taken so far on the journey. What has worked for me. And what I’m still trying to achieve.

Useful links

The $250 Biohack That’s Revolutionizing Life with Diabetes Bloomberg’s August 2018 article with an excellent overview of DIY artificial pancreas technology

OpenAPS Hybrid closed loop system developed by Dana Lewis and Scott Leibrand in collaboration with Ben West and Nightscout

Loop Hybrid closed loop system developed by Nate Racklyeft using the Rileylink developed by Pete Schwamb

EWnxPIVA

Looptips Fantastic information on how to optimise your blood glucose once you start using Loop. Very helpful for people using other looping systems as well.

DIYPS blog by Dana Lewis with fabulous insights into OpenAPS and living with T1D tech

seemycgm blog by Katie DiSimone with all sorts of useful tips on DIY especially Loop

diabettech Where diabetes and technology meet, great website by Tim Street

bionicwookie David Burren’s Aussie looping blog. Lots of fantastic tips and information

KCGM TV Weston Nordgren’s fascinating 2016 Nightscout Skype interview with Ben West about his fundamental contribution to the DIY artificial pancreas

 

Next … Getting ready to loop