Comparing @tim_omer HAPP with @MedtronicUK Smartguard… How do they compare? #gbdoc

I’ve had the 640G for a while now, running SmartGuard, and I really like the technology, as I’ve mentioned in a few blog posts (if you hadn’t noticed), but since I got my hands on the 640G uploader, courtesy of the guys here, I thought it might be quite interesting to run the two alongside each other and see how the two compared, at least on the “low” determination, and insulin suspend option.

To outline the test that I’m undertaking, once again, it’s not terribly scientific. I’m running the following software, and seeing what happens:

  • 640G Uploader – v0.3.0-Snapshot
  • HAPP – code:39 name:1.8
  • NSClient v1_11
  • Xiaomi Redmi 3 Phone running MiUI 6

I’m not undertaking any specific tests to understand what each of the two systems do, this is simply a comparison of what happens when you run them side by side and see the comparison.

Caveats include that I’m not running the latest version of the Uploader, so there are occasions where I lose connectivity, and the target range on the pump is slightly different to that of HAPP as they don’t quite work in the same way, so I’m working to a bit of an estimate.

The other major difference is that I can’t run HAPP all the time as there are occasions where I need to recharge both the phone and the CareLink.

Due to the limitations of local data transfer with the set-up so far, the 640G Uplopader is feeding data to NightScout and I’m using NSClient to retrieve the glucose levels and feed HAPP. This works well enough, although there are occasional hiccups with NSClient and provides a reasonably effective operating model where there is either mobile coverage or Wi-Fi, although I think that local publishing is probably necessary for the Uploader for this to really work safely.

After a few days of use, the anecdotal view that I have can be summarised as follows:

  • HAPP and SmartGuard appear to suspend insulin at more or less the same time. I’ve noticed minor differences in the timing that can be attributed to the sampling time start up, but they seem to be pretty well matched in the “shut down the insulin” stakes.
  • HAPP is more aggressive in its determination of when to resume basal insulin. SmartGuard waits until it sees an uptick in the direction of travel before the basal delivery restarts. HAPP is happy to resume when the levels are flat.
It’s an interesting comparison to make. I’ll run the two of them side by side for longer and see what the outcome is, although the OTG cable doesn’t have the best purchase in the phone, so I tend to have to be static for it to work well.
What’s really noticeable in the first few days of comparison is the difference in IOB. Even with the Insulin duration set slightly longer on HAPP, the IOB curve seems to think that I use insulin more efficiently than the 640G’s curves. Based on my experience, I think the 640G is nearer what I see than HAPP.
My take on the less aggressive resumption of insulin from the 640G is that it’s a commercial product that they want to ensure they are completely avoiding a Hypo. As a result, they are far less aggressive in when they resume basal insulin following a suspend before low.

The reality is that I need to collect quite a lot more data to provide a proper view of the set up, but from an initial comparison, it seems that the low suspend in both is pretty similar. It will be an interesting data comparison to pull all the raw data and compare!

***Updated with HAPP version following comment***


  1. Thanks – post updated with appropriate information. It will be interesting to see how these things play out. I'll be using adjusted basal information from HAPP in nightscout and comparing to Medtronic suspend and restart timings.

  2. Which dosing algorithm are you using inside HAPP? Oref0, or something else? And how low was the BG when it flattened out and/started ticking up? There are some thresholds in oref0 that make it more aggressive for not-very-lows, but keep insulin suspended until BG gets back up above the threshold. Those have changed a few times between oref0 versions.

Leave a Reply to Treets Cancel reply

Your email address will not be published.