How do January’s AI predictions work?

Generative AI

We have spent several years developing our own generative AI to predict a person’s glucose levels. Our machine learning models are built upon millions of data points comprising wearable, demographic and user-reported data from thousands of users.

The output of these models is estimated blood sugar curves.  These models, including their construction and their use in a digital app, are discussed in detail in our peer-reviewed publication.

Personalization

In this app, we use demographic information (height, weight, age, sex, health state) to create prototypical glucose curves for a given person. For a more personalized experience, try the January Pro app, where you’ll use a continuous glucose monitor (CGM) to create a “digital twin” for more precise predictions and robust insights.

Food Impact and Glycemic Index

Virtually all nutrients affect your glucose response in some way, but a few key nutrients contribute the most. Fiber, protein, and healthy fats combine to slow down digestion, which prevents your body from rapidly absorbing sugar. Carbohydrates and sugar have the opposite effect, quickly breaking down into glucose.

Glycemic index (GI) is a measure of how quickly and how much a food will raise your blood glucose. Foods with high GI, like white rice, potatoes, and sugary cereal, are broken down quickly, and will cause higher and faster blood sugar spikes. Foods with low GI, like whole grains and fruits, are broken down more slowly, with more gradual blood sugar spikes. GI is primarily determined by a food’s macronutrient composition (protein, fiber, carbohydrates, fat, and water). You can find more information about GI here, from the Academy of Nutrition and Dietetics.

Traffic Light

January’s rating system classifies foods by their projected glucose impact, simplifying the user experience with a “traffic light” system. We categorize foods as having a low, medium, or high impact based on a user’s self-identified health state (healthy, prediabetes, or type 2 diabetes).

The food impact calculation is determined by the food’s projected glucose impact relative to the user’s fasting baseline. The table below contains information about fasting baseline and spike threshold (the glucose level above which a food is considered to “spike” one’s blood sugar).

The CDC provides guidance around fasting baseline and spike threshold.

Disclaimer

January is for informational purposes only. It does not provide diagnosis, treatment, or medical advice. To learn more about your metabolic health, please consult your doctor, who may suggest an A1c test or a continuous glucose monitor.