Metabolic Digital Twin — Predict Your Glucose Before You Eat

A metabolic digital twin is an AI model of your own metabolism that learns from your glucose, meals, activity and sleep to predict how your body will respond — before you take a bite. Instead of generic diet rules, it gives you a personalised forecast: this meal at this time of day, after this much sleep and activity, is likely to move your glucose this way. It is a screening and decision aid to help you plan, not a medical device.

Why one-size-fits-all nutrition falls short

People respond very differently to the same food. Landmark research on personalised nutrition — notably the Zeevi and colleagues 2015 study of postprandial glycaemic responses — showed that two people can eat an identical meal and see opposite glucose reactions, and that a model built from an individual's own data predicts responses far better than counting carbs alone. The digital twin builds on this idea: your response is yours, and it can be learned.

What the twin learns from

The more it sees, the sharper it gets. The twin draws on the signals that actually shape your glucose:

• Glucose data from a CGM or manual readings — the ground truth it learns against.
• Meals you log, including composition, portion and timing.
• Activity — movement before and after meals changes how food lands.
• Sleep — short or poor sleep tends to raise glucose responses the next day.

From these it builds a personalised picture of your metabolism and keeps updating as your body and habits change.

How the prediction works

When you consider a meal, the twin estimates the likely glucose curve for you specifically — the size and timing of the rise, and how quickly you return to baseline. You can compare options side by side, test the effect of a post-meal walk, or shift a meal earlier in the day to see a flatter response. Over time the forecasts get more accurate because the model is grounded in your own history, not population averages.

What it is — and is not

The digital twin is an educational screening and decision-support aid. It helps you plan and understand, and it works alongside your clinician and your own judgement. This is a screening aid, not a diagnosis — confirm findings with a clinician. It does not diagnose diabetes or any condition, it is not an FDA-cleared diagnostic device, and it does not make medication decisions. Any changes to medication, including insulin or GLP-1 medicines, must be supervised by your clinician.

How people use it

1. Preview meals before eating and pick the option likely to keep you in range.
2. Discover your personal high-impact foods and better-tolerated swaps.
3. Test the effect of timing, sleep and movement on your response.
4. Bring the insights to appointments to guide clinician-supervised changes.

Frequently Asked Questions

Is the metabolic twin a medical device?

No. It is an educational screening and decision-support aid, not a diagnostic device, and it is not a substitute for your clinician. Confirm any findings medically.

Do I need a CGM to use it?

A CGM gives the richest data and the most accurate forecasts, but the concept also works with consistent manual readings. Predictions improve as the twin learns from more of your data.

Is this based on real science?

Yes. The personalised-glycaemic-response approach it builds on is well established in the nutrition-science literature, including Zeevi and colleagues 2015. The twin applies those principles to your individual data.

Can it help reverse type 2 diabetes?

It cannot treat diabetes by itself, but by helping you keep glucose in range through food, timing and movement it can support the lifestyle changes linked to remission in trials like DiRECT and Virta Health. Results vary.