This one helps you find out why.
Your meter records numbers. It cannot tell you whether the walk helped, or how sure anyone should be. This works that out from your own data.
01The difference
A number on its own has nowhere to go. The same number set against every other meal you have logged, with the walks marked, starts to be worth something.
A tracker says
12.2 mmol/L at 21:40
True, and you already knew. It happened, it is written down, and nothing follows from it.
This says
Across 156 meals, walking afterwards was associated with a 1.1 mmol/L lower rise.
Testable. And below it, in full, everything that comparison does not account for.
02One timeline
Glucose, meals, medication, movement and sleep on a single stream, in the order they happened.
Every line says where it came from. A filled mark is something we measured. A hollow one is something we estimated, and it says how confident that estimate is. You should never have to guess which is which.
03A finding
Findings come from a statistical model, not from a language model. A language model may phrase one; it is never allowed to invent one.
And when there is not enough data, the answer is that there is not enough data, with what to log to change that. That is a real answer, so we show it rather than filling the space with something friendlier.
Meals followed by a walk were associated with a 1.1 mmol/L lower rise.
04Testing it
The prediction is written down before the trial begins, in the same moment it starts, and cannot be edited afterwards. When the six days are up, you record what happened and it is scored against that number.
Neither side of the comparison can be revised once written. That is not a promise about how we behave — the database refuses the edit, and refuses to let a trial start without an expectation or finish without a measurement.
Over time that produces something unusual: a record of how often this system was right about you specifically.
Proposed trial
The same breakfast for six days. Walk after three of them, chosen at random.
05Your appointment
You can bring one page instead of a year of screenshots: what changed, what you tried, what the evidence actually supports, and the questions worth asking.
It is built to the standards clinical systems already use, so it stays useful when you change clinic, device, or country.
−1.1 across four measurements
7.9
Nov
7.4
Feb
7.1
May
6.8
Aug
78%
Above target most often between 21:00 and 23:00.
Walking after the evening meal, six days. −1.3 mmol/L against a predicted −1.3, recorded before the trial began.
06Who it is for
Type 1, Type 2 with or without insulin, gestational, prediabetes, or a form your clinician has classified some other way — the record is built for all of it. One timeline, your meter or CGM, meals, medication, activity and labs, the safety rules, and the summary you bring to an appointment.
Findings are narrower. The patterns are found by models built on Type 2 physiology, and they are offered for Type 2 and prediabetes only. For everything else Wellovue keeps your record and says so, rather than running the wrong model over your data and handing you a confident answer you would have no way to check.
That is a limit we would rather state than hide. The panel is generated from the same rule the product uses, so it cannot say something the software does not do.
Everything else — your record, the timeline, medication and labs, the safety rules, and the summary for an appointment — is built for every kind of diabetes and works now.
Type 2, not using insulin
Findings today
Type 2, using insulin
Findings today
Prediabetes
Findings today
Type 1
Recorded, not yet interpreted
Wellovue does not yet produce findings for Type 1 diabetes. The Type 2 analysis is built on a different model of the body and would be misleading here, so it is not used.
Gestational
Recorded, not yet interpreted
Pregnancy changes what these numbers mean, and Wellovue does not yet produce findings for it. This is being built as a clinician-supported workflow.
Other, clinician-classified
Recorded, not yet interpreted
This form of diabetes needs clinician-supported interpretation, which Wellovue does not yet provide.
07Boundaries
This helps you understand patterns and prepare for a conversation with your clinician. It is not a clinician, and it does not pretend to be.
These limits are enforced in the system itself, not promised in the copy. An experiment marked as needing review cannot be started without it.
Bring the readings you already have. A CSV from your meter or CGM is enough to start.