results from Dexcom and xDrip data
Many people have asked questions about the accuracy of xDrip’s algorithm in comparison to the latest Dexcom G4 algorithm, codenamed 505.
To answer this question for himself Andrew Abramowicz decided to get xDrip and an original Nightscout rig to upload data to the same database at the same time, with both sets of data coming from the same Dexcom G4 sensor/transmitter. This then is a true test of how things worked for Andrew and his family. Thanks Andrew for allowing me to use this these images on this page.
NOTE:You can click on most of the images to view the full image in your browser.
Andrew’s son wears a Dexcom G4 CGM sensor and transmitter and for a while has used a Nightscout uploader rig to send CGM data to the cloud to be viewed on a Nightscout web site.
Andrew recently built a xDrip device, which can also upload its data to the cloud to be viewed on a Nightscout web site.
The Dexcom G4 receiver (which is part of the uploader rig) and the xDrip both read the same data from the same Dexcom transmitter and sensor.
Both are loading their data up to the same cloud database which is then linked to the same Nightscout web site.
Data from both is overlayed together, allowing for an easy visual comparison and ultimately to see any discrepancies.
If you’re used to seeing much higher numbers – perhaps you live in the USA? – then don’t panic, the glucose values show are in mmol/l because Andrew is in Canada. To work out a mg/dl value from a mmol/l value just (!) multiply any numbers by 18, so 5mmol/l is 90mg/dl, 10 = 180 etc.
Top left is the time ’10:50′ and showings that the last CGM data received happened 1 minute ago.
Top right is the last CGM reading of 7.2mmol/l (129.6mg/dl) and this has stayed the same from the previous reading.
On the right is the range axis, showing 22mmol/l at the top and 2 at the bottom.
The dots show show the glucose readings, with green dots being actual readings and blue dots being projected readings.
The rightmost green dot is the last reading of 7.2mmol/l from one minute ago.
The rightmost green dot is actually two dots, one for Dexcom, one for xDrip but they are both the same value.
The first reading shown on the left shows that there was a difference between Dexcom and xDrip of approximately 0.4mmol/l (7mg/dl).
The two red dots on the left are where a calibration has taken place, one for Dexcom, one for xDrip.
Here’s an image showing how close xDrip and Dexcom are for the majority of the time.
The next image is of an overnight test showing a hypo in the middle. During the hypo the variance was the largest Andrew has ever seen, before the correction with glucose it looks to me to be about 0.4 mmol/l out, straight after the correction either Dexcom or xDrip appears to have gone wildly out for one reading.
However, no-one I know would rely on CGM data anytime near a hypo situation and never should any treatment been done with first taking a finger prick blood glucose test.
Although there hasn’t been a calibration for 10 hours values are almost exactly the same, maximum out is 0.4mmol/l (7 mg/dl).
The CGM trace below shows a trace over 48 hours – you can scroll left/right.
At times you can see there’s a difference, potentially 1 mmol/l out at maximum point.
Upper line is at 8mmol/l (144mg/dl), lower line is at 4mmol/l (72mg/dl). Red dot indicates a calibration.
Click here if you want to view the full image
Here’s a test placing the xDrip at different distances away from the transmitter, showing that at 25 feet it ability to receive data is impaired, yet at 10 feet it is perfect.