Dexcom CGM needs better rate-of-change visual display

Continuing the discussion from Dexcom Users Direction Arrows:

A recent discussion about the meaning and use of the Dexcom arrows versus the dots suggests that this visual indicator could be improved.

I used to work as a technician in an avionics shop for a major commercial airline. Many of the instruments and cockpit displays integrated a long history of aviation evolution. Many changes to features in cockpit displays were informed by actual airplane crashes! The displays built in the whole field of ergonomics because at 30,000 feet high, life literally hangs in the balance.

The various instruments and displays must communicate a strong visual presentation dense with vital information. They pilot cannot visually fixate on one display at the expense of focusing on myriad other systems and what appears outside his/her cockpit window. Situational awareness suffers too much complexity and confusion.

Why re-invent the wheel?

Therefore the various instruments and displays that my shop serviced were well honed devices that benefitted from decades of use. Their visual presentation was concise, clear, intuitive, and right to the point.

One of the instruments that my shop serviced was the Vertical Speed Indicator or VSI. This analog indicator consisted of a pointer mounted on a round display that deflected up or down to immediately indicate whether the aircraft was gaining, losing, or maintaining altitude.

A flight level indication was a pointer holding steady at the 9:00 oā€™clock or pointing left condition. If the pointer moved clockwise, or up, it meant the aircraft was climbing. If the pointer moved counter-clockwise or down it meant the aircraft was descending. One quick look and the pilot could immediately learn this important flight info.

Hereā€™s a visual of what this instrument looks like:

Itā€™s calibrated in 100ā€™s of feet per minute. Intuitive, simple, and fast to read.

Could medical device builders learn from commercial aviation?

I suggest that the Dexcom (and other mfgrā€™s.) could borrow this design concept and build into our CGM displays a much better way to represent the speed of blood glucose movement up or down. The current system is misleading. The linked discussion shows the confusion it sows relating to the all too frequent disagreement between the arrow and the dots.

When the Dexcom arrow is level, it only means that the blood glucose levels are not moving up or down by more than 1 mg/dl for every minute. In other words a level arrow can mean anything from truly level blood glucose to one moving up or down 15 mg/dl every 15 minutes. The latter condition is anything but level in my book!

CGM manufacturers could easily incorporate the intuitive smarts built into an aircraft VSI. Now I know that there is a about a 15 minute lag in the interstitial indication of blood glucose when compared to a fingerstick reading. Weā€™ll have to live with that reality for the time being. But the current situation could be markedly improved by building in a presentation informed by commercial aircraft VSI display.

A CGM display that informs instead of confuses

The up or down markings on an analogous CGM display could mark off +1, +2, +5 and +10mg/dl per minute as well as the -1. -2, -5, and -10 mg/dl per minute.

I also know the the We Are Not Waiting movement has been able to use the Dexcom ā€œraw dataā€ for their benefit. Perhaps this raw data is more frequently available than every 5 minutes. That would enable CGM VSI to track more closely and update more frequently.

This doesnā€™t seem like a complicated upgrade to the sometimes misleading arrow system and it communicates actionable info to the brain of the person with diabetes. And it may help pierce the confusion that the fog of high and low blood sugar that impair us sometimes.

Does anyone else think this is a good idea?

3 Likes

Not much. The screen could say bg now 125, bg in ten minutes will be 140 without intervention. :slight_smile: I think looking at the dots is as easy as observing a fine distinction in arrow angle unless the arrow was long. If the up arrow rotated from nine to two for a fast rise it would not read intuitively.

@AARON10, look at the VSI picture. Imagine a 1 mg/dl (per minute) marking where the ā€œ5ā€, up or down is depicted. The arrow would pass through 5 minor divisions or hash marks before it got to the 1 mg/dl per minute rate of change. All we get now is a sideways arrow and dots that suggest it is something less or more than level. There were more than one remark in the original thread that noted that the arrows was all that gathered their attention.

Not sure what you intended by this comment. Are you suggesting that the Dex CGM offer a prediction like the Medtronic 640 pump that suspends insulin delivery based on predicted lows?

Yes, the Dex uses a predictive algorithm so the 15 minute interstitial fluid delay is compensated for. So what is the point of the arrow. Itā€™s a prediction as to what will happen. Itā€™s like the car dash lighted gas pump picture I see when my tank is low. I understood the graphic that you tried to describe I think. Itā€™s an arrow that goes up and down and over in more detail than the popular arrow system of flat, diagonal, vertical, two parallel vertical arrows. And the arrows average the vertical deltas over a period of several dots. Is your idea to eliminate the averaging and so instead change needle defection with each dot incorporating only the one dot period? I agree I donā€™t like seeing a diagonal arrow when the graph is flat, but that happens when there is a bend in the curve. Maybe the display could say ā€œchange in curve, flex pointā€ thatā€™s all Iā€™ve got for now but you did a lot of thinking and creative design and are to be commended for trying to help.

Your first comment was unclear to me. I thought you didnā€™t understand the concept I was trying to convey. It appears you do and donā€™t think itā€™s a good idea. Thatā€™s fine with me.

The point of the arrow as I propose is to provide finer, and I think better, granular detail to the data in between each state of the current Dex arrows. Itā€™s kind of like zooming in on a Google map so you can read the street names and see which ones are one-way. To me itā€™s the difference between an analog engine temperature gauge on my car and the ā€œidiot lightā€ when the car overheats. The current flat arrow meaning spans all the way from a 15 mg/dl over 15 minute climb or descent. Thatā€™s not flat and misleads. Now, Iā€™m savvy enough to look at the dots for finer info but why obscure this data?

I hear you but I disagree. I have no illusions this idea will influence design going forward but one can hope. Itā€™s just a reflection on my past experience and a wishful story. Perhaps TuD is not the right readership for this. Maybe the community over at CGM in the cloud would be more fertile ground.

Thank you for your thoughtful reply.

This would be a great option, giving more information to those who want it. I was excited when I read about the partnership between Dexcom and Google, but disappointed that it wasnā€™t to improve the receiver display, which could offer SO much more than it now does. Two obvious ones to me:

  1. Be able to change the vertical scale. I try, somewhat successfully, to live with a good flatline, which means most of the time those inflection points and changes in rate are almost impossible to see unless I bring the receiver right up to my face.
  2. Have an outdoor setting for things like exercising so you can glance quickly at one or two giant, bright numbers in strong sunshine.
    Alternative displays like the altimeter analog would be great to have as options. And since weā€™re on the topic of improvements, I sure do wish they could make the receiver even a little bit water-resistent, but I know thatā€™s a different thread here.
1 Like

Would you spell out that cgm cloud location for me please? I promise not to be a Debby downer there. What did you think of my concern about ā€œintuitiveā€ I didnā€™t expand my thoughts yet. I think there would be a learning curve. Pilots are trained. And the altimeters are less prone to error than cgm individual readings. Also to conserve screen space maybe we could have just the arrow with how many minutes,seconds itā€™s been at that angle? Sort if Iā€™m not being helpful I keep thinking the graph is the best. On minimedā€™s system you could back up the display showing the actual numbers. I missed that at first but because of the generalized nature, not to say lack of certitude of the numbers and because being within a ten points estimate is good until one tries to aim for 83 and itā€™s the slope that is of most importance. now Iā€™m getting less serious Iā€™d like to see the integral under the curve and the second and third derivative. But we donā€™t want a tangent line, which is the instantaneous slope we want a 15 minute average. And any display will have that averaging error. A tangent line could be alarming, non intuitive because of non-altimeter type fluctuations.

I whole-heartedly agree! You should be able to zoom and cut out that 200-400 mg/dl waste of real estate. If that was allowed it would make those inflection points easier to see.

I also agree with improving the readability in bright sunshine. Iā€™ve complained for years about that also in the pump displays. I fear that we are such as small niche market that the economics of integrating these changes are prohibitive.

If my idea were implemented, I donā€™t see it as a second screen display; it could easily make use of the sizable wasted space on the current screen.

Thank you for your comment.

The CGM in the Cloud group is on FaceBook.

I donā€™t get your reference to ā€œintuitive.ā€ That term appears nowhere in your earlier comments.

The aircraft instrument that I used as an example is not an altimeter. Perhaps that is confusing your understanding of my idea. An altimeter just reads altitude. A Vertical Speed Indicator reads rate of change in altitude. Itā€™s this rate of change feature that the CGM display shares.

Iā€™m not proposing replacing the graph with my concept. Your comments make me feel like we are two ships in the night sliding past each other with no communication connection.

What does this mean?

I am not as facile with terms like ā€œintegral under the curve,ā€ "second and third derivative, and ā€œtangent.ā€ Perhaps you could explain so that others reading this (including me!) can understand your thought process.

Last word of first post.

Itā€™s the rate of change of the altimeter and the rate of change of the blood glucose reading. Iā€™m not sure in what way the blood glucose readings and the altitude are not the subject. for example say the bg changes but no it was an error. Now say the altitude changes but no it was an error. The first one happens the second one does not happen.

Sort should read sorry. For the very thing you noticed.

What the run on sentence meant was:
I missed the ability to see the number associated with each dot when I dropped elite cgm. But I recognise there is a lack of accuracy so a graphical display showing relative distance is better than the false accuracy of numbers. Although if your goal is 83 mg/dL, the lack of accuracy is catastrophic. And when oneā€™s bg is 100 or 90 or 110 which may all have the same result on the com, that is to say be indistinguishable, what one cares about is the slope, is it steady, rising, or falling and how fast.

The tangent is the first derivative. It is the rate of change during one second, the instantaneous rate of change, not the slope between to points. The slope between two points on the curved graph is the average rate of change, not the instantaneous rate of change. The second derivative is how fast the needle is moving when every second has itā€™s own rate of change. The third derivative is how much the needle movement is slowing down or speeding up. The integral is the area between the graph and something else, letā€™s say 120.

Unfortunately, as far as I know, ā€˜raw dataā€™ is not available more frequently. Transmitter is sending ā€˜raw dataā€™ every 5 minutes. Receiver filters the raw data some more, and obtains a new value every 5 minutes. The filtered data is what we see on the receiver screen (and this is what is shared via Bluetooth on G4/Share). In the Nightscout/We Are Not Waiting project, people have figured out how to extract and display both the raw and the filtered data. Seeing raw data can be useful in some cases when there is significant noise from the sensor. Raw data is also available during the 2-hour warm-up period for a new sensor. Dexcom considers such data unreliable, which is why the receiver does not display anything during the warm-up period.