“data smoothing” is not the same as the more detailed signal processing that the G6 does, and always has. (The so-called “data smoothing” in older versions of the G6 app was solely about adjusting prior readings that varied from the otherwise more natural curve.)
The explanation for all this is detailed in my substack article that explains how glucose works in the body, and how the G6’s signal processing differed from the G7. In short, glucose is not evenly distributed in the bloodstream, and tends to clump together in fluids. When you measure any given sample of fluid that contains glucose, it will vary from the next drop, etc. That variability is more dramatic when glucose levels are higher. And even more so when glucose levels change. It’s all about the biokinetics of glucose.
When you try to measure glucose in complex fluids, like blood or interstitial fluid, the variability is even more dramatic.
To truly assess (or guess) what a person’s systemic glucose levels are, you need to gather a continuous series of samples and apply some math to figure out that real number. This is not merely a matter of averaging numbers! One must understand and apply laws of physics that apply to glucose molecules within fluids.
The guy that came up with the algorithm that translated the CGM signals to produce glucose values for the G6 was an expert in this area, which is why people who used the G6 were able to achieve much better glycemic control than other sensors.
The G7’s algorithms abandoned that method (for reasons discussed in my article), but the rumor mill on various discussion forums seems to agree that it was mostly spawned by manufacturers of automated pumps, who wanted to apply their own algorithms and not rely on Dexcom’s. So, the G7 no longer applies these algorithms, and simply reports a very raw, rudimentary series of glucose levels that happen to reflect actual glucose values from various samples. Yes, these are “accurate,” but only if you compare in individual samples. As such, what the g7 reveals is the more realistic volatility of glucose erratically floating around in samples.
There are many business reasons why this is good for Dexcom (think “data licensing” for all those pump manufacturers), as well as claiming at the G7 has a better MARD than the G6, but these benefits may be short-sighted if the poorer performance of the G7 causes a user backlash. One cannot predict if this will ever come to pass.
If the G6 were ever discontinued, it would be a shame for all those T1Ds who rely on it to achieve healthier outcomes.
