My AI doppelganger tells me what to eat
A journalist spent about 100 days using Twin Health’s AI-powered “digital twin”, which uses health data to predict how her glucose levels may respond to food and activity. The experience shows how digital twins could support personalised metabolic advice, though the technology remains experimental.
The programme provided a continuous glucose monitor, smart scale and blood pressure cuff, and analysed blood tests alongside Apple Watch data. The app colour-codes food choices and updates its predictions as the user logs meals and glucose responses; in one example, protein Cheerios with milk rated orange alone but green with a walk afterwards. Twin Health is currently available through some corporate or private health plans, generally for people with chronic metabolic issues.
- A digital twin predicted how food and activity might affect glucose.
- The journalist logged health data daily for about 100 days.
- Twin Health remains experimental and limited to certain health plans.
New here? Start with this
A digital twin is a computer model built from information about a person’s body and habits. Twin Health’s version uses health measurements and daily activity to estimate how food and exercise may affect blood sugar, then adjusts its advice as more data comes in.
The programme combines readings from devices such as a glucose monitor, scale and blood pressure cuff with blood tests and smartwatch data. It is aimed mainly at people with ongoing metabolic health problems and is offered through some employers and private health plans.
The idea is to tailor health guidance to each person rather than rely only on broad dietary rules. These systems are still being developed, so their predictions and the benefits of following their advice remain under study.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
AI digital twins could make metabolic advice more personal and practical by combining glucose readings, blood tests and activity data, helping people see how everyday choices affect them. For people with chronic metabolic conditions, that feedback may support healthier habits and give clinicians useful information, while wider availability could make tailored guidance easier to access.
The case against
The advice may appear more precise than the experimental predictions warrant, and colour-coded food ratings could oversimplify nutrition or encourage anxiety and restrictive eating. Continuous monitoring also raises questions about privacy, cost and unequal access, while results from one person’s experience do not establish that the programme improves health outcomes.