Basal Testing for the 670G Auto Mode

I started up auto this morning and have a question going in. This has been on my mind for a year. As Auto is a “dynamic insulin delivery” system, does that mean its “basal” delivery is exempt from basal testing? Every hour will be different in auto mode, so there would be no way to replicate a 4 hour window, but I would still like to be able to do some basal testing to see how effective it is… if that makes sense. I can’t use it to adjust anything, but I would like to use it to gauge performance.

Does anybody have any thoughts on this? Or ideas on how to run such a test? Ultimately, I’d like to do an overnight one, but I’d be happy to start anywhere. I can go out and find basics on basal testing and get it going on my own, but I wanted to ask whether or not the same rules should apply… And I’ll even accept mansplaining… I’ll just pretend like it’s a woman saying it. :smiley:


You have a good idea of your existing MDI basal, correct? Why not start with that minus 10%, then leave the delivery rate constant overnight, then tweak until it looks flat with no IOB? For us, we use the night hours after 2am to validate a good basal.

1 Like

Actually it does not make sense to me in regards to the 670g.

And is it even accurate that “Every hour will be different in auto mode …” or is it more like every minute is different?

I can’t possibly pretend to understand the 670g basal in auto mode. At this point, my suspicion is that nobody actually understands it. The best that one can hope for is to have it work for one’s self.

1 Like

Is there a data download from 670? Like Carelink, or whatever it was.
That might give you insight.

Or the endo or CDE may have access to tools that us dumb patients don’t get access too.

This is simple enough except there’s no way to do this in Auto Mode. I had originally asked the question to find out whether or not I should be able to test Auto Mode’s “basal” delivery the same way as I would any basal. I see @Thomas has answered that he doesn’t think so. I’d like to understand why not?

Yes, somewhat. We were in the process of tweaking.

I’ll use this as soon as I’m back in manual. Shamefully, I’ve never really participated in the assignment of my pump basal rates more than to carry out the testing given to me by an educator. This will give me something to look forward to once I’m done with auto. Thank you!

1 Like

What is the reason you can not do as @Michel has mentioned?

Okay… and why not?

I’ll tell you where the question originated. I started wondering this after a debate with someone about Auto Mode basal delivery. I argued that I wasn’t getting enough insulin compared to my manual mode basal rates. They argued that my manual mode rates were wrong. Which would be true if they were. But my blood sugar is fine in manual. It’s not in auto. It got me thinking that I could just run a basal test in auto to see if it keeps me even the way I remain in manual. Yes, every 5 minutes is a freshly calculated micro bolus, but why would the expectation that they would accumulate to provide enough insulin to remain even change??

I’m not arguing with you in the least. I’m trying to learn what’s in your head. :grin:

Hence my previous question.

Carelink doesn’t reflect the size of the micros…

Maybe I didn’t understand you. How would this relate to basal testing in auto?


Perhaps I simply misunderstand.

Can you briefly explain what you mean by “basal testing in Auto”.

Not “how”.

But a brief description of what that concept means to you.

1 Like

“When evaluating basal insulin levels, keep this important fact in mind: In the absence of food, exercise and mealtime/bolus insulin, basal insulin should hold the blood sugar steady. Any significant rise or fall in blood sugar during a fasting basal test probably means that the basal rates need to be adjusted “

So this… This is true of pumping as well as of MDI, obviously… But this pertains to predetermined rates. Is there any reason Auto mode’s “dynamic delivery” exempts it from having to pass the same test?

My brain is mush right now. I am trying to communicate something, but I’m not sure even I understand it. :woman_facepalming:

I still don’t understand what “basal testing in Auto” means to you.

My guess is what you mean is you want to know what it does. But that is only a guess.

1 Like

I thought maybe you were thinking historical basal. So if had run in auto for 2 days, downloaded data, it would somehow show what was delivered in auto mode that was not relative to micro bolus given related to carb BG rise.

1 Like

I’m going to gather myself, @Thomas and @MM2 and then try to return to clarify. I was trying to answer things last night when I should’ve been sleeping.

I think at the core of all of this is a desire to make sense of it… what auto is doing, and after a second frustrating evening, a second frustrating morning, and really my best efforts at improving my numbers, I’m trying to get my mind wrapped around the fact I might not be able to satisfy that, And even if I could ultimately do that, at what cost would it have occurred? I am feeling pretty frustrated this morning, just a little irritated, and I think I’m taking a mental health day.

MM2, you suggested a regroup session. I haven’t had the time to sit and do it, which both a symptom and perpetuation of the problem. I can’t even remember what you recommended doing… was it to list everything that I need to tend to? Or just reconsider priorities??

Today feels like sinking. :woman_facepalming:

That was my impression.

So I think a good place to start might be:
“Is it realistic to expect to be able to get an answer?”

If “YES” then drive on.

If “NO” then perhaps redirect?

In any event only my opinion.

1 Like

Yes, you communicated this last night. It got me thinking about it.

Diplomatic to a flaw, I have a hard time with yes OR no. It’s yes AND no. The last time I was in, I was able to give it an enormous amount of attention. I didn’t have the Facebook group, and I didn’t have you guys. It was on me to figure it out, and it’s what got all of my time after what I spent with family. I don’t have that time available now. Not even close. At this time, I don’t have the bandwidth to try to understand what auto is going. Being in it without understanding it is a waste of time.

So then, no.

You’re kind to respond. Having a bad day. WAS having a bad day. It’s been upgraded to a “not very good” day.


I think that is valuable right there.

A lot of people have tried to figure this out. @drbbennett put a lot of work into it and has lots of posts on it. At this point, I do not think it is reasonably possible to understand HOW it works. IMHO that is a waste of effort and will only lead to frustration.


I do not think that means it is unusable. You CAN use something without understanding HOW it works. Most people who drive cars probably don’t actually understand how the engine works. They just trust it.

If the auto mode on the pump can be turned on and let it go and if it provides results that you consider acceptable then perhaps it is not necessary to know the HOW of the internals.

Some of the people who have studied this (@drbbennett) have shown a few ways (which sound less than intuitively obvious) in which you can provide a level of control over the auto. However you (apparently) can NOT FULLY CONTROL it. That is simply how it was designed. By all appearances. Auto is auto. If it works FOR YOU that is great.

The reality is that there likely is not a single pump on the market anywhere which would work for everybody. Some pumps work for some people. The 670g [in auto mode] appears to work for some people and not work for others.


@Nickyghaleb, I think you came up with the right kind of idea above, however it was mixed in with some other ideas. Your comments got me thinking and this is what I would do.

If you would like to do an experiment I would do the following:

Put yourself in manual mode set your basal, do your basal test, i.e. find a time where you can safely go 4 hours or so without food and see what happens. If flat for those hours, then great, calculate how much insulin was given for those 4 hours. If not flat, keep adjusting until you get a flat response. Then use that amount of insulin.

Put yourself in Auto mode, do the same thing, i.e. don’t eat, and calculate how much insulin is given when you are flat for the 4 hours.

Compare and contrast.

1 Like

You are very capable of improving your numbers. The solution just might not include auto mode. It is not a failure to accept that auto mode doesn’t give you the control you want. It’s basic design and fail-safe features limit it from working the way you want it to.


In addition to the excellent advice given above my @MM2, I would say that trying to understand how a black box works is basically worthless, because you can’t control what the black box does.

If it isn’t obvious, it goes without saying that the 670 uses a LOT of black box functionality in it’s auto mode. You don’t need to understand it, you just need to determine whether it works for you or not. Since by definition you don’t have the ability to adjust the black box programming.