GlucoBeat
Log in
Type 1 diabetes

With type 1, every meal is read with the insulin next to it

GlucoBeat puts every meal, every dose, every workout and every night on the same timeline as your sensor, measures what happened next and tells you when insulin or exercise are mixed into the response. It never suggests doses.

Why it fits type 1

In type 1 the response to a meal depends on the dose, the timing and the insulin on board. GlucoBeat doesn't hide that: it flags it.

Insulin, as an event

You log each dose (rapid or basal) with one tap. On each meal you see if there was a dose nearby, and the response is flagged as “other events nearby” instead of being attributed to the meal alone.

Exercise and the following night

Each session with its kind and intensity, what changed during the session and whether there was any low in the next 24 hours, compared with days without exercise.

Your usual response, with caveats

Repeated meals with median, range and n, and how many of those times were clean of nearby insulin or exercise: confidence takes it into account.

AI analysis that asks, not orders

Patterns in your data, written as associations, and concrete questions to take to your endocrinologist. Never doses, ratios or correction factors.

What it looks like in the app

A day with its doses and exercise on the curve, the sessions by kind and intensity with what happened in the next 24 hours, and a real, complete AI analysis. Example data except for the analysis.

A day with GlucoBeat

Example
Your glucose Target range 70180 Sleep

Exercise by kind and intensity

Example
ExerciseSessionsDuring the sessionLow in the next 24 h
Walking · moderate8−14 mg/dL1 of 8
Running · intense4+9 mg/dL2 of 4
Padel · intense3−22 mg/dL2 of 3

Reference: on days without exercise there was a low on 2 of 14 days. Associations in your data, not causes.

Analyse how you're doing

Real example

Pick a period and the AI goes through your readings and everything you've logged to tell you what keeps repeating.

Everything you need to understand yourself

AI analysis

Pick some dates and get concrete patterns: what doesn't suit you, what does, when you tend to go low.

Nothing slips through

If you forget to log a meal, GlucoBeat spots the rise it left in your glucose curve and suggests the approximate time. The app asks you about it: all you do is say what it was.

Statistics that matter

Time in range, variability, GMI, lows, AGP profile and comparison with the previous period.

Experiments

What happens if you go for a walk after dinner? {n} different comparisons: log each situation a few times and GlucoBeat compares with your own data, with its n and its confidence.

Your usual response

The same dish, seven times: median rise, usual range, time to peak and what changed the times you walked afterwards. With its n and its confidence.

How would this affect me?

Type what you fancy: first what happened in your diary with similar meals and, if you want, an AI estimate with a range and what tends to make a difference.

What a repeated meal looks like

With insulin nearby in 3 of the 5 times, confidence drops and it says so. Associations in your data, not causes.

Margherita pizza

Example, made-up figures
Times logged
5
Median rise
+72 mg/dL
Usual range
+51 to +98
Peak at
95–140 min
No other events nearby
2 of 5
Confidence
Low

What GlucoBeat does not do

  • It never suggests, calculates or adjusts doses, ratios or correction factors, neither through the AI nor through any calculation. What you log about insulin is there to read your data, not to decide.
  • It is not a medical device: it does not diagnose, give doses or change any treatment. That is your care team's job.
  • It does not alert on values and does not work in real time: your sensor and its app do that. GlucoBeat shows you the evolution and what goes with it.
  • It does not predict: every figure describes what happened in your data, with how many times it was seen, in what range and with what confidence.

Frequently asked questions

Does it take insulin on board into account?

It takes it into account as an event: if there was a dose near a meal, the response is flagged as not attributable to the meal alone and confidence drops. It does not calculate insulin on board or use it to recommend anything.

Does it work with a pump?

Yes. You log boluses as insulin doses and the rest works the same. The pump's basal does not need logging.

What about hypos?

Every low stays as an episode in the statistics, and treating a low is logged as its own event. GlucoBeat does not alert on lows: your sensor does.

How much does it cost?

The diary, charts, statistics, usual response, report and sharing are free forever. The paid plans widen the limits, such as those of the AI or experiments.

Start with this week's diary

Bring in your sensor readings and log meals and doses for a few days. Free and no card.

Start for free

FreeStyle, Libre, LibreView and LibreLinkUp are trademarks of Abbott; GlucoBeat is not affiliated with Abbott.