The Research Behind SNAQ

Most people have no idea how far off their carb estimates are. Research shows people typically miss by 20 to 30 grams per meal when estimating manually. That kind of gap is hard to spot in the moment but shows up clearly in your blood sugar afterwards.

Researchers at Bern University Hospital tested SNAQ's photo-based food analysis on 48 real meals, 128 food items, including breakfasts, cooked meals, and snacks. They compared SNAQ's estimates to laboratory-verified values using a precision scale. The mean absolute error for carbohydrates was 5.5 grams.

SNAQ also measured well for other macronutrients: 1.3g mean error for fat, 2.4g for protein, 41.2 kcal for energy. Average processing time was 22.9 seconds per meal.

Earlier computer vision systems reported carbohydrate errors of 14.8 to 26.2 grams. SNAQ's approach showed meaningfully better accuracy across that range.

Source: Herzig D, Nakas CT, Stalder J, et al. Volumetric Food Quantification Using Computer Vision on a Depth-Sensing Smartphone. JMIR mHealth and uHealth. 2020;8(3):e15294.

Knowing SNAQ is accurate in a lab is useful. Knowing how it holds up against real people counting carbs every day, and against another leading app, is a different kind of evidence.
‍
Researchers tested 53 adults with Type 1 diabetes on 26 hospital-prepared meals served in three different portion sizes. Participants first estimated the carb content on their own, without any assistance. They then used SNAQ and Calorie Mama to estimate the same meals.
‍
SNAQ's mean absolute error was 13.1 grams. Manual estimation by people with Type 1 diabetes came in at 21.0 grams. Calorie Mama came in at 24.0 grams. For people whose own carb estimates tend to run high, low, or inconsistently, the gap is likely even more meaningful in practice.
‍
Note on independence: One co-author of this study serves as a consultant for SNAQ. Meal reference values were independently verified using recipe databases and measured weights.

‍Source: Baumgartner M, Kuhn C, Nakas CT, Herzig D, Bally L. Carbohydrate Estimation Accuracy of Two Commercially Available Smartphone Applications vs Estimation by Individuals With Type 1 Diabetes: A Comparative Study. Journal of Diabetes Science and Technology. 2025. PMID: 39058316

Knowing that SNAQ estimates carbs accurately is useful. Knowing that using it actually improves blood sugar control is a different thing entirely.

A randomised controlled trial tested SNAQ against standard carb counting methods in 44 adults with Type 1 diabetes using automated insulin delivery systems. Participants were randomly assigned to either use SNAQ or continue their usual approach for three weeks. The trial was funded by the European Foundation for the Study of Diabetes, independently of SNAQ.

Participants using SNAQ saw a 6.6 percentage point improvement in Time-in-Range compared to the control group (95% CI 2.85 to 10.28%, p less than 0.001). The result was statistically significant. The study also showed improvements in time above range, mean glucose, and glucose variability, all in favour of SNAQ.

This is the only randomised controlled trial evaluating a photo-based food tracking app for glycaemic outcomes in people with Type 1 diabetes.

‍Source: Piazza CD, Kastrati L, Bally L, Herzig D, Nakas C. Efficacy of an image-based automated food analysis app in AID users with type 1 diabetes on glucose control: a randomised controlled trial. Presented at EASD. Trial registration: NCT05671679.

Clinical accuracy is one thing. Whether people actually stick with an app long enough to benefit from it is another. Most food logging behaviour drops off within the first few weeks.

Internal data shows 42% of SNAQ users were still actively logging meals three months after starting. For a food tracking app, that is unusually high.

Internal data, Q4 2023. Based on users who logged at least one meal in month three.