Study Finds Popular AI Calorie-Tracking Apps Underestimate Meal Energy by About a Third

Photo-based calorie-tracking apps, which use artificial intelligence to identify foods in a meal photo and estimate portion sizes, have become a common tool for people trying to manage their weight or monitor their diet.

New research presented at NUTRITION 2026, the flagship annual meeting of the American Society for Nutrition, found that 4 widely used apps in this category consistently underestimated the calorie and fat content of meals when tested against a tightly controlled reference standard.¹

"Photo-based calorie tracking apps are very popular, especially for people trying to manage their health or lose weight," said Aaron Hengist, PhD, a postdoctoral visiting fellow with the intramural program of the National Institute of Diabetes and Digestive and Kidney Diseases, part of the National Institutes of Health. "However, the accuracy of many of these apps has not been thoroughly evaluated. Our study helps address this question by looking at whether these apps can reliably estimate calories."

How Was the Accuracy of These Calorie-Tracking Apps Actually Tested?

The research grew out of a larger diet study at the National Institutes of Health Clinical Center examining how the body processes nutrients on a low-carbohydrate, or ketogenic, diet compared with a standard diet. Because meals for that study are prepared in a metabolic kitchen, a research kitchen where ingredients are weighed to the nearest tenth of a gram, the researchers had access to a precise reference standard against which to test the apps.

Standardized photos of 102 of these meals were run through MyFitnessPal, LoseIt!, CalAI, and Appediet, and each app's calorie and macronutrient estimates were then compared against the meals' actual, precisely measured composition. Olivia Charles, a postbaccalaureate intramural research training fellow at the institute, presented the findings during the meeting's President's Oral Session.

How Does This Compare With What's Already Known About Dietary Self-Monitoring Accuracy?

Concerns about the accuracy of diet-tracking tools predate the introduction of AI-powered features. A 2020 validation study comparing MyFitnessPal's food database against a standardized reference, the Belgian Nubel food composition database, found strong correlations for energy intake and macronutrients, though accuracy was notably weaker for cholesterol and sodium.²

The new research adds to that body of evidence by testing a newer generation of apps that rely on photo-based, AI-driven estimation rather than manual food entry, and by using a considerably more precise reference standard, a metabolic kitchen, than most prior validation studies have had access to.

What Did the Researchers Find About Where These Apps Struggle Most?

The underestimation was not evenly distributed across meal types or macronutrients. All 4 apps estimated carbohydrate content more consistently than fat or protein, and two of the apps, MyFitnessPal and LoseIt!, were more accurate when estimating higher-calorie meals than lower-calorie ones.

In a secondary analysis of more than 200 additional meals, the researchers found that the apps struggled more with meals from a low-carbohydrate ketogenic diet, an effect the researchers attributed to the higher fat content in those meals being consistently underestimated. "People using a photo-based tracking app without adjusting the portions or entering the amounts of food should take the results with a grain of salt," Hengist said. "These apps tend to underestimate calories, especially from fats, so what they actually ate is likely higher than what the app shows."

What Are the Limitations of These Findings?

These results were presented as a conference abstract and, per the American Society for Nutrition's own guidance, have not undergone the peer review process required for publication in a scientific journal; the findings should be considered preliminary until formally published.

The analysis was also limited to 4 specific apps and meals drawn from a single research diet study, so the degree to which these findings generalize to other tracking apps or more varied, real-world meal patterns remains unclear. The researchers noted that combining photo-based estimation with more traditional dietary assessment methods could improve real-world accuracy, an approach that has not yet been tested directly in this study.

References

1. American Society for Nutrition. Photo-based calorie-tracking apps may underestimate energy in meals. EurekAlert! July 25, 2026. Accessed July 30, 2026. https://www.eurekalert.org/news-releases/1136415

2. Evenepoel C, Clevers E, Deroover L, Van Loo W, Matthys C, Verbeke K. Accuracy of nutrient calculations using the consumer-focused online app MyFitnessPal: validation study. J Med Internet Res. 2020;22(10):e18237. doi:10.2196/18237