Medtronic 670 G approved 6 months early

I’m confused about something. I understand that if you are trending low it will shut off your basal, but I know that if I am low turning off my basal won’t matter. I still need a snack. I don’t see that being a huge benefit. I guess my real question is what if you are running high, does the AP system give you a bolus to correct your high?

I guess I’m not understanding why this is a big deal? What am I missing

I think this is where the predictive low program can be effective. If the 670G program predicts a low one hour in advance, then turning off the basal has chance to effectively mitigate the subsequent low. Perhaps instead of going low for 60 minutes at a depth of 45 mg/dL, the system’s response trims that to a 30 minute low at a 60 mg/dL depth, then that would be a better outcome. I’m speculating here since I have no experience with this rig but I think that’s the basic idea.

This basically predicts your High and Low trends, by as much as 30 minutes in advance. So, if the system sees you’re going to cross your threshold (or a pre-determined level above your threshold) in a half hour, it will start decreasing the basal insulin accordingly. Same logic on the High side.

Of course, this predictive tech is only as good as the accuracy of the sensor… so hopefully, this Enlite 3/Guardian 3 sensor is up to the task.


Hi, Mike. Does the 670G also increase basal rates in anticipation of expected highs?

1 Like

yes it works similar to openAPs, issuing high temp basals if a predicted high is coming down the pike.

There’s more ability to this system than I first understood. Adjusting basal rates all day long, up, down, and off is a larger menu of possible actions.


It does, yes.

MedT says there are two modes – Manual and Auto.

When you’re in Auto, it will automatically adjust insulin to push you toward 120 mg/dL (or the temp of 150 mg/dL, if you have that set). So, it will increase basal if you’re above 120 to get you back to that target.

In Manual mode, you can set it to detect when you’re roughly 20 points from your Low threshold, and so it predicts you’re going to cross that threshold and it will allow you to adjust insulin from there.

Still some work you have to do with this “hybrid,” but it’s more automated and in the early stages of an AP than we’ve seen before.


To answer the original question as it was posted: yes, I am planning to get the 670G.

In fact I was just in the process of getting an upgrade to the 640 in a month (my insurance allows upgrades only after the 4-year warranty of each pump I buy ends) and when I read the announcement of the 670 approval had a brief moment of hesitation what exactly to do (proceed with my plan or wait until 670 becomes available). Well, Medtronic just solved my dilemma: – now I am definitely planning to get the 640 as planned before the end of this year and then upgrade again as soon as the 670 is actually on the market.

(For context, while I did demos with other pumps before I decided on my first, I have been a loyal Medtronic customer since I started wearing a pump and CGM (3 months after being diagnosed in 2008) and while the earlier versions of their sensors were in fact painful and inaccurate, I am quite happy with the improvements the company has made over the years. As others have noted, their customer service is second to none. I consistently have an A1c in the 5s with minimal hypo episodes and (for now, knock on wood) no complications. My current and only additional diabetes-related goal is to lower the variation in my glucose levels (currently SD averaging 30-40) and I hope having the additional tools of the new 670 system will help me in that.)

Thanks, Chris, for explaining this for us.

Not a problem. If you have any additional questions, I would be glad to answer. I got a PhD in chemistry building glucose sensors in the 90’s. And as irony would have it, now have a 13 year old T1 son.



This is my comprehensive coverage over on DiabetesMine:

I am, by the way, comfortable referring to this as an Early or Pre Artificial Pancreas.


Mike – I just finished reading your review at DiabetesMIne. It’s a great summary of what this pump does, what it is, and what it isn’t. I think this unexpectedly quick FDA approval bodes well for the speed of D-tech rollout going forward. I can’t help but think that the FDA has gained a new sense of respect for patients in this regulator/manufacturer/patient dynamic. The “we are not waiting” movement also played a role, I suspect.

If this model had a user-selectable BG target, I would be getting in line for it. Your article explains why that didn’t happen and I haven’t read that anywhere else. It makes sense even though I wish it weren’t so. Nice piece.

1 Like

With all the technology available now, I’m amazed that Medtronic chose to not incorporate any method of connecting the pump to a smart phone and smart watch. Dexcom has embraced this idea and there are a multitude of apps that support their products. I just don’t get this backwards thinking.

1 Like

Regulatory compliance… Dexcom isn’t an insulin pump so they can integrate with whatever they want.

It’s also a matter of this 670G is at least a few years old as to design, and since they released the Minimed Connect for data-viewing and sharing, they did not redesign it to include the necessary Bluetooth.

1 Like

This is the key. How accurate is the Enlite 3 compared to the Dex G5?

Wonderful to see tech being pushed forward. Hoping to see an OmniPod/Dex collaboration in the near future.

Nice write up @Michael_Hoskins


Really hoping this new Guardian 3 sensor is up to the task. The “clean” clinical data shows it’s quite a bit better than their Enlite, but not as good as G5. Still, it’s better than G4 original, and if that translates just as well to the real world settings, I think that this puts MedT into the game as a real player with a damn good sensor. We shall see, of course!

Mike - are you moving to the 670G now that you’ve tried it out?

I haven’t tried it, and probably won’t. No, I am not interested for two main reasons: It doesn’t have data sharing, and the 120 isn’t adjustable. Also not convinced of CGM advancements. I’m doing fine on my #PumpHiatus for now. TBD what I do once we get into early next year…

1 Like

I won’t have much a choice either way. United has moved to Medtronic as it’s preferred pump. My pump warranty is up in March, if this new 670G has not started shipping by then…I will wait until it does. Both of my last pumps have been from Medtronic, and I truly have been very happy with their customer service, and quality of the pumps.

No, I am sure it is not perfect. But I am excited about where this is going. I think about all the advancements in CGMS technology the last few years, and I think this will only get better. It’s really the only option available.

I am sure the current 120mg/dl will be tightened as they progress with the technology. I am sure the sensor has to be even more accurate…I’m ok with that, it’ll get there.

I honestly can’t wait to try it. I haven’t felt a smidge excited in a new product the last 8 years…this one, kind of does.