Monday, 14 September 2015

Predicting raw Dexcom values. Part 1

I've been playing around with the "raw" values that can be extracted from the Dexcom G4 sensor. The xDrip android app saves the values to a database which can be exported and inspected. The raw and calculated values are saved, together with all the blood glucose calibration values. To get a better idea of what I'm talking about, have a look at the graph below. The red shows the raw values and the green shows the calibrated calculated glucose values. The blue dots are blood glucose calibration points (This is where I manually checked my blood glucose with a finger prick). Also, remember that glucose measured from interstitial fluid lags behind blood glucose with about 5-15 minutes. It can be seen in the graph at several locations where the blue dot shows a risen glucose level before the graph does.
The current calibration algorithm in xDrip simply fits a y=mx+c line through the calibration points. There is also an age adjusted raw value that is used instead of the standard raw which somehow improves the accuracy slightly. I'm still trying out different calibration algorithms and will report on them at some stage. 
There is also a prediction algorithm currently implemented in xDrip that will show you the predicted glucose value while you wait for the next Dexcom packet to appear. The algorithm simply fits a y=ax^2+bx+c parabola to the latest 3 readings. When predicting more than 7 minutes ahead, the predicted value stays fixed. The parabola can quickly shoot off to a very low or high value, therefore the prediction is limited to a maximum of 7 minutes ahead. The prediction algorithm can be seen in action in the animated graph below. The red line shows the raw values, the green shows the predicted value 5,10,15 and 20 minutes ahead. The animation cycles through 5-20 minutes prediction time. The perfect prediction algorithm should show the red graph in green just shifted to the left 5-20 minutes.

Why is it useful to predict 5-20 minutes ahead?
  • Interstitial fluid lags behind, predicting ahead might help reducing the latency.
  • Knowing what the near future values will be, can improve the accuracy of a calibration algorithm. Especially when the user is trying to calibrate when the glucose levels is falling or rising rapidly.
  • Closing the loop. Artificial Pancreas might find it useful to know what is going to happen in order to inject insulin or glucagon at the right time. 
  • Alerting the user of an incoming low or high before it happens.
I used a basic neural network, nothing fancy, in order to try and predict raw values more accurately than the parabola method. The data was split into 80% training data and 20% testing data. The latest 50 minutes of raw data is provided as the input to the network and the output is the predicted raw value 5,10,15 and 20 minutes ahead. If there was any missing values within the 50 minutes, a fifth order polynomial is fitted to the available values and missing values are estimated by using the polynomial. The animated prediction results can be seen in the image below:
The mean absolute error for the described methods on the testing set can be seen in the table below:
5 minutes 10 minutes 15 minutes 20 minutes
Parabola 4.09 7.03 9.59 12.21
Neural Net 2.55 5.17 8.05 10.41

It does not make sense to predict further ahead unless information such as insulin and carb intake is available. I have more advanced algorithms that will hopefully help improving prediction and calibration in the future. Stay tuned for Part 2.

Moving on to sensor 3 and the backup Tic-Tac xBridge2

I'm already on sensor number 3. Sensor 1 lasted 30 days, Sensor 2 lasted 19 days and I'm currently on day 9 with Sensor 3. When you live with T1D, everyday is a different day and certainly a gift. You don't know how the day will go, the best approach is to take it day by day because everyday comes with it's own challenges. The day on the left was a good day, my glucose stayed amazingly good between my target range (except for a "short" low period shown in red). The day on the right looks like a typical day, lots of ups and downs with a bad high at the end. My 2nd sensor started to show a lot of noise at day 18 which can be seen in one of the images. Surely, time to put on a new sensor when you see stuff like that going on.

I wanted to have a backup xbridge device and I finally got hold of another wixel. Unfortunately, the wixel was already assembled with the header pins on... in order to fit everything into a tic tac box, the pins had to be removed. It caused me a bit of a headache because I'm not a soldering / electronics expert. The easiest way I found to remove the pins was to first cut off all the headers with a wire cutter and then try to de-solder the pins I wanted to use. Placing the tip of the soldering iron on a pin and then pushing the pin out with a needle seemed to work nicely. I followed the plan for the xBridge2 which can be found in the apps/xBridge2 folder from this repository: https://github.com/jstevensog/wixel-sdk. The xBridge2 differs from the first version in the following ways:

  • Requires you to solder 1 extra wire between the BLE module and wixel.
  • Transmits the message between the bridge and xDrip app in binary instead of text.
  • Powers down the BLE module while waiting for the next packet to arrive, saving you precious battery power.
I will be using the xBridge2 as a backup and development device while adding new features to the xDrip android application. 


Wednesday, 19 August 2015

Dexcom G4 sensor 1 lasted for 30 days!

I had a sensor stuck in my arm for 30 days!!! It was time to remove my first Dexcom G4 sensor. I could probably have used it for a couple of more days, but I started to get more and more noise from the sensor. In the image on the right you can see the graph did not look very smooth any more. The low glucose alarm falsely woke me up a couple of night, because of random noisy low values. It lasted 30 days!! I'm extremely happy with the results. I would have been happy if it lasted only 14 days. The sensor stayed stuck nicely the first 7 days, only then did I add tegaderm over it. After 14 days I added another piece of tegaderm. The sensor seemed a bit loose on day 21, but it was still functioning well. I placed a sock over it for the last 7 days, which helped a lot to keep it in tack and it looked much neater. My Dexcom reader is also not used any more, I don't carry it around and rarely looked at it. The xDrip application is just working so much better than the reader. I'm wearing the sensor at the back of my upper arm. It was easy to limit the amount of water getting to the sensor. The only downside I had with wearing it on my arm was that I experienced a few compression lows during the night. A compression low is when the sensor is pushed against the body and it reports a false low glucose value. Typically this sometimes happen when you sleep on top of the sensor.

My current sensor configuration can be seen in the image. I'm covering the white part of the sensor sticky tape with a tegaderm that contains a hole such that the transmitter fits in the hole. I also added another larger tegaderm over the sensor and transmitter. The nice thing about tegaderm is that you can hardly see it and it does a good job in holding things down with the additional benefit of being waterproof. The black thing around my arm is a sock cut open. I keep my xDrip bridge device inside this sock. The transmitter reads the glucose levels from the sensor. The xDrip bridge collected the signals transmitted by the transmitter and sends the data to my cellphone via bluetooth. My xDrip bridge also buffers 40 minutes worth of data, so I don't always need to be near my cellphone. Obviously I take off my xDrip bridge when I shower or sleep, but the sensor and transmitter stays in all the time.

Tuesday, 4 August 2015

Climbing the T1D mountains

I've been woken up by my Dexcom / xDrip system several nights in a row because of a low. Waking up every 2 hours and having to eat something is not very pleasant. Last night I decided things must change and I tried something different and it failed terribly. On the left, my glucose graph from around 20:00. On the right, a popular mountain... can you spot some similarities ?? At 21:00 my levels started to rise, I did not want to try and fix that with exercise and gave myself a small correction bolus. At around 23:15 I really wanted to go and sleep, but noticed that I over-corrected, going low. Having several nights with many lows, it is definitely not a good idea to go to sleep already low. I ate a few pieces of malva pudding, thinking that I'm going to make sure I don't get a low during the night.... I ate too much (very little actually, but the pudding can mess with your glucose levels). 01:00, I woke up realizing I'm climbing a massive mountain and it looks like it's going up for-ever. Injecting another correction bolus and going back to sleep. Another mistake I made was to switch off my high glucose alarms, otherwise I would have noticed the peak much sooner. The correction bolus was not enough and the dawn phenomenon kicked in at 05:00, creating another peak. Getting out of bed at around 06:30, realizing I'm on a mountain and quickly gave myself a correction and breakfast bolus. If you fall from a mountain, you fall hard, at 08:00 I was crashing fast. I was driving to work at that time. I consumed anything I could find to eat in my vehicle, I knew that I had to try and stop the rate of the fall as much as possible. Everything takes time, food takes time to digest, and then there is momentum, it is difficult to catch a falling egg without breaking it. I managed to turn the curve to go up just in time. Then the oscillation starts again, at 10:00 I'm going over my target range again because of all the things I had to eat to prevent the fall. Staircase exercise to the rescue, and I'm down going low again at 11:00. Luckily the fall was not that high this time and 100g of yoghurt quickly corrected and stabilised the oscillation. Sometimes it is very difficult to manage the T1D mountains. I lift up my eyes toward the T1D mountains, from where will my help come? My help is from the Lord, maker of heaven and earth. (Ps. 121)

Wednesday, 29 July 2015

Gaming my glucose levels with FreeStyle Libre and Dexcom G4


FreeStyle Libre vs Dexcom G4

The day I started using my Dexcom G4 system, I still had about 3 days left on my last FreeStyle Libre sensor. That gave me the opportunity to take images such as this one seen on the right (Dexcom reader at the top, Libre reader at the bottom). Having experience with both, I'll list some random thoughts on each.

Freestyle Libre

  • Sensor insertion process was very easy, I like the round flat shape of the sensor. They use awesome glue that easily stay stuck for 14 days.
  • I don't like the fact that you HAVE TO remove it after 14 days, it stops working after 14 days.
  • Having a blood glucose meter inside the reader is very useful.
  • It is nice having access to reports quickly directly on the reader such as : Average glucose, time in target, graphs of previous days.
  • You can't zoom into the graph, it's difficult to look at exact measurements, you can't really see them on the device. Exporting the data only gives you a value every 15 minutes, no raw 1 minute data available.
  • No alarms. Many times I scanned my glucose and I only then realised that I was going high while I could have prevented the high if I was notified by an alarm a few minutes earlier.
  • You don't need to carry the reader with you all the time, the sensor will keep on storing data (only the most recent 8 hours worth of data).
Dexcom G4
  • Sensor insertion process a bit more complex, but not too bad.
  • You can restart the sensor after 7 days. You can wear the sensor until it becomes inaccurate or falls off. I'm currently on day 12 of my first sensor and the FreeStyle Libre's glue is much stronger. I have applied tegaderm to keep the sensor in position.
  • It asks you for calibration values every 12 hours, so you are forced to verify the accuracy of the CGM. Note that you should only calibrate within the normal ranges and when your levels are stable. Don't calibrate under extreme conditions, the maths will get confused and you will mess things up.
  • I can get the sensors locally and because you can go longer than 7 days, it is much cheaper than the Libre.
  • It does have alarms, you can customize it somewhat. It woke me up a couple of nights and a couple of times during the same night. You can not disable the emergency super low glucose alarm, I guess it makes sense that you should not ignore that alarm.
  • The Dexcom receiver does not show any reports or statistics... although the Dexcom studio windows application do.
  • The Dexcom receiver have a latency of 10+ minutes... clearly you can see that they are filtering things a lot... don't worry, all is not lost, xDrip to the rescue..
  • My Dexcom receiver is not using the 505 software update, the update is only available in the US and is suppose to fix the latency issues.
  • Now for the awesome DIY stuff:
    • I've build the xDrip bridge device and I'm running xDrip on my cellphone. It means that: The xDrip bridge collects data from the Dexcom G4 Transmitter and transmits it to my cellphone via BLE (Low energy Bluetooth). These are "raw" values, prior to any calibration and filtering done on the Dexcom receiver. I don't even need the Dexcom receiver any more. 
    • The xDrip application is open source. I can modify it as I like.
    • The xDrip application contains features such as: reports, statistics, customizable alarms, zoom-able graphs, rate of change information, time till next reading, etc..
    • xDrip can also upload your data to a database in the cloud. The Nightscout project allows you then to monitor the glucose values remotely from any device that can open a website.
The Libre sensor and Dexcom G4 transmitter seen below:

Gaming my glucose

Managing my T1D turned into a game. The more I use a CGM, the more obsessed I become to manage my values as well as possible. My score consists of the following measurements:
  • How close can I get my average to 5.0 ?
  • How small can I get my standard deviation ?
  • How long can I keep the levels stable between 4.0 and 8.0 ?
  • Ultimately, the blood A1C will tell me somewhat how well I did in the past 3 months.
Insulin works pretty slowly, even the rapid acting insulin. I had to try and improve somehow on getting my level back to normal when going high. Giving a correction bolus usually takes 1-2 hours to bring back my levels to normal range. I tried some light exercise, and it is working very nicely. Whenever I see my levels going up and beyond 8.0, I try to find stairs. Climbing stairs for about 5-10 minutes dramatically improves the rate at which the levels go down. It takes about 10-20 minutes for the levels to drop instead of 60 minutes (when taking a shot). Taking a correction bolus can also cause a low if the amount of insulin was too much. Having a low can cause a high if you eat too much in order to correct the low... can you see the oscillating cycle ?? Doing something else rather than taking a correction bolus definitely have some benefits. On the right you can see the xDrip android application with a staircase scenario. I only started climbing stairs at the 4th orange dot, note the rapid decline after just a little exercise. You can also see the oscillating effect on the 24 hour graph at the bottom of the image which happened because of my breakfast bolus being too much. Just note that a T1D still needs insulin, exercise and diet alone can not control it.

My target range is set to 4.0-8.0 mmol/L. I've been on the Libre for 70 days and my time in target was 59%. I've been on the Dexcom G4 for 11 days and my time in target was 73%. It might be that the Dexcom alarms made the difference, or I'm just learning how to play the game better. Either way, thanks to Abbott and Dexcom, this game is very difficult to play without a CGM. Libre on the left, xDrip with Dexcom G4 on the right.



DIY !!!

The one thing I don't like about the Dexcom G4 is that the reader or xDrip bridge needs to be nearby (a few meters max) all the time otherwise you will miss values. When you try playing this game, you would like to see all of the data and don't loose any values. The xDrip bridge listens for 2.4GHz packets and when it receives one, retransmits it via BLE to the cellphone. After that, the xDrip sleeps for about 4.5 minutes before it starts listening for a message again. If your cellphone is not in range, you will loose the packet. This means that you have to carry the xDrip bridge and cellphone with you everywhere. I want a little more freedom. I made some changes to the wixel c code running on the xDrip and java code running on the xDrip android app. I was using the downloadable xDrip apk on my cellphone, which means that I would loose my data if I had to start using my modified application. I did not want to loose my data, so I had to add some code to fix that.  My branch of the code is at https://github.com/JacoCronje/wixel-xDrip and https://github.com/JacoCronje/xDrip-Experimental. My list of modifications:
  • Firstly, I did not want to loose my data. The android app already had an export function, but no import function. I've added the import function which I used to import my data after removing the installed xDrip version and installing my version.
  • xDrip wixel stores the raw glucose values and packet id in a circular buffer. There is still plenty of RAM left on the wixel. I can fit 2 days worth of values in there, but I'm storing only 20 minutes worth of data in the buffer for now. The dexcom packet id contains a byte size value that increments with 3-5 every reading. I'm guessing that it's some sort of timer value and I'm using it to determine the time-stamp of previous readings relative to the latest reading.
  • The BLE packet only transports 20 bytes at a time. I had to change the xDrip android code to combine packets in order to receive messages larger than 20 bytes. I've added a starting([) and ending(]) symbol to each packet. The android code then use those symbols to determine when the complete message arrived. 
  • Code that ensure that historical values don't get added more than once to the database.
In the future I would probably add more logging functionality. I like to keep track of when I injected what kind of insulin and how much. I'm not so much a carb counter, but that might also be useful. Currently I'm using nightscout to log the information, but it will be much better if it can be done directly in the xDrip android app. 

Tuesday, 21 July 2015

Updated T1D survival kit


My latest T1D survival kit can be seen in the image. See the list below with links to the websites where applicable:
  • A: Freestyle Optium blood glucose monitor.
  • B: Finger pricking tool
  • C: Needles for finger pricking tool
  • D: Blood glucose test strips, you need a strip for every blood glucose finger prick test. 
  • E: Ketone test strips, in case of high glucose levels, you need to test for ketones. If detected, emergency measures need to be taken or you could die. 
  • F: Apidra insulin, rapid acting insulin, required to inject before I eat. 
  • G: Lantus insulin, slow acting insulin, I take two injections daily. 
  • H: BD micro fine needles, attach them to inject the Apidra and Lantus. 
  • I: Glucagen emergency hypo kit. If I become unconscious because of low glucose, give this to me. 
  • J: Dischem glucose sweets and my favourite chomp chocolate, needed to counter hypo's.
  • K: Alcohol wipes to clean the skin before inserting a CGM sensor.
  • L: FreeStyle Libre Reader. The reader also have a blood glucose monitor and you can use FreeStyle optium test strips.
  • M: FreeStyle Libre sensors. These are used and don't work any more.
  • N: Dexcom G4 Sensor. 
  • O: Dexcom G4 Transmitter. This attach to the sensor. You can buy the Dexcom products from the local distributor Ethitech.
  • P: Dexcom G4 Receiver, this receive the signals from the transmitter and displays it on the screen. The device also alerts you when going low or high.
  • Q: My xDrip bridge device. DIY project. The bridge receives the signals from the transmitter and re-transmit them via bluetooth to my cellphone. The xDrip android app then uploads the data to a database and can be viewed remotely from anywhere. If you created a xDrip device, you no longer need the Dexcom receiver. The xDrip method seems to be much better, you get the raw values which is less fudged than the values displayed on the receiver.
  • R: Nightscout, used for displaying and alerting remotely on any device that can open a website. You need N, O and Q for this to work. 


Wednesday, 15 July 2015

xDrip progress

I have started putting my xDrip components together over the weekend. It took me less than 10 minutes to solder and assemble all the parts, and I'm really not into soldering, anyone can do it. However, next time I will use shorter wires, remove the pins from the BLE module and use a smaller charger module. I'm going to use this configuration for now until the size of it gets annoying. The wixel code was also super easy to compile and upload. How does it work ? In short, the wixel reads the signals coming from the Dexcom transmitter and sends that to the BLE module. You pair your cellphone with the BLE module and the android app then receive the data from the BLE module. The android app displays the glucose values and can then upload the data to a database. The data can then be viewed remotely by multiple people simultaneously all over the world. I already received my Dexcom kit from Ethitech and eagerly wait for my last Freestyle Libre sensor to expire. Less than 4 days away and I will be part of CGM in the cloud and the NigthScout project. #WeAreNotWaiting !!

Monday, 6 July 2015

Sensor 5 inserted and getting ready for xDrip

My stock of Libre sensors are depleted, I just inserted my 5th and final sensor. My 4th sensor lasted the 14 days without any problems. Here you can see my epic 5th sensor, sucking into my arm.

For the people still wondering how the device looks that inserts these sensors: The left image shows the device with the sensor still intact before insertion. You simply press the spring loaded device against your skin which will then release the spring. The needle that you can see in the centre will penetrate the skin, leaving the thin measurement wire inside. The right image shows the device after insertion.

What will I do after sensor 5 ?? I discovered that Ethitech is a distributor of Dexcom products in South-Africa. I can buy a Dexcom G4 sensor for about R1100 from them. Officially the Dexcom sensor is only approved to be used for 7 days, but many many people have shown that they still work after a few weeks. Now, if the Freestyle Libre was available easily in South-Africa, it would have been more difficult to decide how to continue. The only way I could buy a Libre sensor was through ebay, ... they are selling a sensor for more than double the retail price... I'm not willing to pay R2000+ for one sensor and take the risk of loosing it via the postal service any more.

Getting ready for my xDrip!! The xDrip is a cool DIY device that reads the signals transmitted by the Dexcom transmitter. The signals are then packed and send over Bluetooth v4 (BLE). A cellphone can then read the bluetooth packets. After that, many cool things are possible. Data can be uploaded with the cellphone onto the internet and viewed remotely by someone else. Prediction algorithms can be written, alarms can go off when glucose are going low or high, my custom algorithms might improve the MARD accuracy, location information can be used to inform relatives where the patient is when going low, .. and many more. I received all the parts needed to assemble my first xDrip today. I've got 2 weeks to get it working which should be more than enough time, I only have something like 6 wires to solder and software to upload to the wixel.



Tuesday, 23 June 2015

Sensor (1-3) Performance

It was time to change sensors again, starting my fourth sensor today. I had some accuracy issues with my 3rd sensor, so I decided to have a closer look at the data. From the exported .csv file, I wrote a small piece of software that find pairs of blood glucose and Libre scanned values. Normally I scan with the Libre just before or after I take my BG, so extracting the data was not a big issue. Ideally, the BG and Libre values should match almost perfectly. The performance of these kind of sensors are usually measured by the Mean-Absolute-Relative-Difference (MARD). Abbott claims that the Freestyle Libre have a MARD of 11.8% after the 3rd day. The lower the value, the better.

The BG values can be seen in the x-axis, while the y-axis shows the Libre values. For a perfect sensor and BG combination, all the points should be on the y=x line. My first sensor results shown in red, second sensor in green and third in blue.

The following plot shows the values for all 14 days for each sensor. (click on it for a larger view).
Clearly sensor 3 seems to always measure a bit low. Values below the line indicates that the Libre measured a value lower than the BG value, the inverse applies to values above the line. Sensor 2 seems to be measuring little too low in the [2,8] range, but too high in the [8,16] range. Sensor 1 measured a little too high on most cases. Abbott should really add the option to calibrate. The device already have a BG meter build-in, so why not add an optional calibration / offset process ???

The following plot shows the values for 13 days, skipping day 1 for each sensor.
You can see that most of the values from day 1 are outliers when comparing the two plots.

The MARD values can be seen in the following table:
Day 1-14 Day 2-14 Day 3-14 Day 4-14 Day 1-7 Day 8-14
Sensor 1 14.7% 15.0% 15.3% 16.2% 14.5% 15.0%
Sensor 2 12.1% 12.0% 12.6% 12.6% 13.2% 10.9%
Sensor 3 18.4% 14.7% 13.3% 12.1% 20.26% 11.8%
This shows that the accuracy do improve over time. Although sensor 3 was looking very bad in the early days, the accuracy quickly improved and overall performed better than sensor 1.

The mean error can be seen in the table below:
Day 1-7 Day 8-14 Day 1-14 Day 2-7
Sensor 1 -0.20 -0.68 -0.36 -0.30
Sensor 2 -0.07 0.15 0.03 0.30
Sensor 3 1.39 0.58 1.21 1.05
The mean error table clearly shows that sensor 1 was giving values slightly high and sensor 3 gave very low values, a simple calibration process could have improved the accuracy a lot.

Let's assume a simple calibration procedure: After 7 days, take the mean error (of day 2-7) and use that as a simple calibration offset. If we use the mean error as the offset (100%) or half of the mean error as the offset (50%), we get the following MARD from day 8-14: (0% no calibration)
100% 50% 0%
Sensor 1 12.4% 13.4% 15.0%
Sensor 2 9.17% 10.0% 10.9%
Sensor 3 12.7% 9.68% 11.8%

The system can definitely get some benefit from a calibration process.

Thursday, 11 June 2015

Issues with new sensor and the Dawn Phenomenon

It was time to remove my second sensor a few days ago, and time to start up my third sensor. Image taken right after I let my wife pull off the sensor. Again, no issues with keeping it stuck for 14-15 days. The extra tegaderm over the sensor do help and you can see the ripple effect it caused on my skin.


The dawn phenomenon could be clearly seen in action during many nights. Around 4am, my glucose levels suddenly starts to rise. This is due to the liver releasing glucose in order to prepare me for the day (If I understand it correctly). The graphs below show the sudden rise (+-4am) while I was still sleeping. It gets kind of difficult to manage this, you need to perfectly time the long acting insulin to peak at 4am or start the night on a lowish glucose level (which increase the probability of a hypo during the night), or just make sure to wake up and eat breakfast asap after 5am. I really like my sleep, getting up early is not my favourite option.



My 3rd sensor is giving me some issues. It seems to be always lower than my BG readings (offset issue). There also seems to be some scaling issue. Abbott is going for a calibration free option with the FreeStyle Libre, but it would have been great if they just added an option to calibrate the sensor in case something goes wrong or is a bit off. The readings still follow my BG trend, but needs an adjustment.
Due to my limited stock of sensors and no Libre support in South-Africa, I'm keeping the sensor in, hoping that the offset might improve over time. The asterisks (*) on the graph shows the BG readings (done with a finger prick test). 







Yes, I'm not having a good BG day today, but the readings from the Libre seems to improve a little, hopefully it will be as good as the previous two sensors soon.







Tuesday, 26 May 2015

Goodbye sensor one, Hello sensor two


















My first sensor easily lasted the 14 days, it was still stuck pretty well with only some of the edges looking a bit scruffy. It feels kind of sad to take it off, but that is how life goes. Pulling it off revealed a lot of sticky glue. Initially I was worried that the sensor can easily just fall off, but after seeing how the glue/adhesive look after 14 days, it made me think differently. It will take more than just a little bump to accidentally loose one of these.



This is how the sensor looks on the "other" side. You can see the small needle looking thing in the middle. It is flexible and not very hard. When you insert the sensor, a larger sharp and strong needle is used to insert the flexible one. A spring loaded mechanism is used to quickly push and pull the needle in and out. The hole that you see in the sensor from the outer side is there in order for the larger needle to be pulled back. Cleverly designed.


I applied some skin-prep before inserting the new sensor. It was again a painless and quick exercise of inserting the new sensor, no blood seen this time. I had a bad night's sleep, so I don't expect my levels to be awesome within the first 24 hours and most likely this post contains many errors.




Friday, 22 May 2015

11 Days with a sensor still stuck in my arm.

Time flies when you are continuously monitoring your glucose levels, almost time to remove my first sensor! The sensor is still stuck and working nicely, although some of the adhesive tape on the edges seem to be lifting. I kept tegaderm on the sensor just in case it decides to get loose. Sometimes it gets slightly itchy around the sensor, nothing major yet, it's going to be interesting to see how it looks beneath the sensor when I have to remove it.

The snapshot report of the past 11 days shows some interesting statistics that I was not able to capture before. After reducing my long-acting insulin by about 17%, I still have some issues with hypo's early in the mornings. This might be due to Lantus peaking at around 4-6 hours after injection. The amount of information available is just amazing, I feel much more in control of my BG levels. Look at my average number of scans / day ... 57 !! Imagine pricking your finger 57 times a day! Impossible, you will be using a whole pack of test strips each day. The FreeStyle Libre simply gives you the freedom to check your glucose as often as you like.


When you have been using only BG finger pricking tests for 9 years, any CGM report looks pretty awesome.  The weekly summary reports provide nice detail about what happened when. I experienced some low's and high's during the night, which I was unaware of. I can see the benefits of having something like the Dexcom G4 that will alert you when such things happen. I'm just wondering why did they not put a small speaker inside each Libre Sensor... just a simple beep would have been sufficient.
On the 17th of May I ate too much after a hypo and it caused my levels to be way too high through-out the night. That error pushed my average for the day up to 9.4, not acceptable at all.
I also realised how much glucose coke contains... in the early mornings of 18 May I had a hypo and only sipped 2-3 sips of coke, you can clearly see the spike that it caused. Yes, I needed that spike at that stage, but I can just imagine what happens to the poor non-T1D that drinks many glasses or bottles of coke.


On the 20th of May I opened a new insulin pen.... and my levels were just not going down as expected. You can see that I injected a total of 39 units (8 shots) on that day, normally I inject about 20 units. At the end of the day I decided to throw away the insulin pen, and open another one. I immediately got much better reaction from the new pen, this was not the first time it happened to me.
I have an update on my 3rd ebay parcel... somehow it was send back to the sender and it never left Germany, so he is mailing it again. If it happens again, I'm asking for a refund. At least the South-African Post Office can not be blamed this time. Currently it's such a pain and expensive to get the sensors in the country, chances of me switching to Dexcom is increasing as you read this. I have tasted the benefits of having a CGM and I'm afraid I won't be able to go back from where I started.

Thursday, 14 May 2015

My FreeStyle Libre experience. Day 1-3.

I waited about 16 hours after inserting the sensor before activating it. According to other users waiting 12-24 hours have an effect on the accuracy of the sensor. I had no issues with sensor accuracy yet, right  from the start it was remarkably accurate. The sensor is very comfortable, no pain at all. I'm almost completely unaware of it.

Thus far, I'm very impressed with the FreeStyle Libre, it will be difficult to go back using only a BG meter. If Abbott does not release the Libre in SA by the end of my sensors, I'll most likely switch to the Dexcom system. You can't really manage T1D without a CGM. The amount of extra information available is just awesome! Just the advantage of checking your levels any-time is great and I almost felt guilty of checking the levels so often, it feels like you are consuming/using valuable resources (test strips?) each time, but actually the information you get is for "free".

Before I begin with my analysis, I'll just give a quick summary of how I treat my T1D. My pancreas does not generate insulin at all, so I have to inject myself with insulin in order to manage my blood glucose (BG) levels. I don't use an insulin pump, I use multiple insulin injections. My BG goes up when I eat and goes down when I inject insulin. Exercise will lower my BG, but it is not enough to control my BG, I need insulin. My body slowly release stuff throughout the day and night that increases my BG slowly, Lantus (Long acting insulin) are used to cover this BG rise. Apidra (Rapid acting insulin) are used to cover my BG when I eat. The black line in the graph shows the insulin release curve of my rapid acting insulin. The red line shows the release curve of the long acting insulin. So, it becomes a complex system to control and balance. The amount of insulin injected must match the amount and type of food eaten, exercise done and body state (sickness and other factors also influence BG levels). I don't do carb counting, it just seems too complex to exactly determine what is in the food you eat, I adjust my insulin amounts by learning from experience.

What have I learned from Day 1 ?


Below you can see the graphs generated by the FreeStyle Libre software, you can export everything to a .csv file which is very handy. The values with asterisks are BG levels tested with blood (test strips), this clearly shows that the Libre is accurate. The other values (shown with small circles on the graph) were collected when I scanned my BG with the Libre.


The spike at 10:00 is because I ate two rusks with my coffee.. I usually have to do this, otherwise I'll hypo soon after 10:00. Without the spike, the downhill from 10:00 - 12:00 will let me crash with a hypo. The question now is, what caused the downhill from 10-12 ?? It can't be the rapid acting insulin that I injected at 7am, because most of that will already be depleted (remember the release curve graph). It must be the long acting insulin, I'm injecting too much of it! Through-out the day my BG is dropping and I have to prevent hypo's by eating snacks at regular intervals. Have a look at Day 2 below, very similar pattern, eating myself out of hypo's all the time.
Lesson learned on Day 1: I need to reduce the amount of long-acting insulin.

What have I learned from Day 2 ?


Wow, look at that hypo during the night. I woke up at 3am and tested my BG. It was low. I've been low for about 3 hours! Without the Libre, I wouldn't have known that. The spike with my two rusks can again be seen at 10-12. This time a little bit lower because I did not add one teaspoon of sugar to my coffee. Remarkable what only one teaspoon of sugar can do! Same pattern of lowering through-out the day can be seen.
Lessons learned on Day 2: a) I'm having long hypo's during the night. b) One teaspoon of sugar is bad for your BG levels.

What have I learned from Day 3 ?


I reduced my long acting insulin by about 17% (4 units). I had another very long hypo during the night and did not sleep very well, resulting in struggling to maintain a good BG level around 9-11. Overall the reduced long acting insulin seems to work better, we will see how tonight goes.
Lessons learned on Day 3: a) I need to stop those night-time lows. b) When you don't sleep well, your BG levels will go up during the day. c) You can trust the accuracy of the Libre.


A summary of your low-glucose events (hypo's) are given in the report. Average duration of 100 minutes during the night !!! Life-saving information gained within 3 days of use, I don't need to say more.

Monday, 11 May 2015

11 May is FreeStyle Libre Day !!


Yes !! I finally have a FreeStyle Libre in my hands. The DHL express parcel containing the reader and 3 sensors were delivered this morning. I also received an unexpected call from the Johannesburg International Mail Centre (JIMC), they found my "missing" parcel ! I drove all the way to pick it up from them, not taking any more risks of them trying to deliver it to me. So, now I have 5x sensors and a reader, still one parcel missing, but I can at least get going with what I have received. 

My BG meter is a FreeStyle Optium meter, which turns out to be very beneficial. The test strips are compatible with the FreeStyle Libre and the reports generated by the software also show the results of your manual BG checks. I'll be using my Libre for normal BG checks as well as CGM style checks when I do have a working sensor on me. My initial thoughts on the report generating software is that it can be very very useful. 
Inserting the sensor was painless, quick and easy. The needle looked a bit scary, but did not hurt at all. Let's hope the needle and sensor stay stuck for the next 14 days. I got some tegaderm and vet tape from Dischem to apply in case the sensor needs extra adhesive. I will try to be patient and wait for 12-24 hours before activating the sensor. I could see some blood coming through the centre part of the sensor, so it makes sense to let the wound heal a bit before activating the sensor. 



Thursday, 7 May 2015

Trying to get a FreeStyle Libre

After reading about the FreeStyle Libre, I obviously wanted one very badly. So do many other people. Abbott launched the Libre in Europe, not yet in the USA (approval in progress) and sadly not yet in Africa. The demand is so high, that new customers in Europe have to go on a waiting list which seems to be at least 3 months long. My initial idea was to let a friend in Europe try to buy it for me from Abbott and then we could arrange for an African delivery... the original plan failed due to the huge demand.

My next step was to jump onto ebay and try to see if I can get the device and a few sensors somehow. Luckily you can find it on ebay, but it comes with a price, the sensors easily sell for double the original price. I took the risk and bought a starter pack (device + 2 sensors) and a few sensors.. Some of them from UK sellers and some of them from Germans. The UK items were shipped to my friend in the UK, who shipped them to me via DHL express. The German parcels were shipped separately via DHL. 

Now, let's have a look at the status of these parcels.

The first German parcel safely departed from Germany,.... and then vanished... not sure where it is now ??
The second German parcel safely arrived in SA, cleared customs and EMS claimed that they couldn't find me in 2009 ?? Now the item is on hold somewhere. I phoned them every day this week, after several hours someone answered and they will "look" for my parcel, or they will "deliver" it tomorrow, or iesh! I still don't have it :(

The UK parcel, still looking good! On track and on time! Let's hope I'm lucky this 3rd time... just one problem, it's not in SA yet, the tracking fun might start when it arrives in SA soon.