Accuracy of ChatGPT , Gemini, Claude and DeepSeek in Carbohydrate Counting
Diabetes, Obesity and Metabolism · 2026
DOI: 10.1111/dom.70747 ↗Kurz gefasst
Fachartikel in der Zeitschrift Diabetes, Obesity and Metabolism, 2026. Verfasst von Luca Zagaroli und 7 weiteren Autor:innen. Beteiligte Einrichtung: Unit of Pediatrics, San Salvatore Hospital, L'Aquila, Italy. Hauptthemen laut MEDLINE: Artificial Intelligence, Diabetes Mellitus, Type 1, Dietary Carbohydrates. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 43 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Aims
To evaluate the accuracy of four general‐purpose artificial intelligence (AI) models—ChatGPT (OpenAI), Gemini (Google), Claude (Anthropic) and DeepSeek (DeepSeek AI)—in calculating the carbohydrate content of meals compared with clinicians‐calculated reference values. Materials and Methods The primary endpoint was equivalence between clinicians and AI‐generated calculations within an error margin of ±5%. One‐hundred twenty‐four meals were analysed, equally distributed among breakfast, lunch, dinner and snacks. Carbohydrate contents were jointly determined by two paediatric diabetologists and one clinical nutritionist using the USDA FoodData Central and CREA Italian Food Composition Tables. Each AI model received identical, standardized prompts in English describing the meals. Statistical analyses included the Two One‐Sided Tests procedure, the Bland–Altman plots, the Wilcoxon signed‐rank and the Spearman correlations.
Results
The clinicians' median carbohydrate content was 30.32 g. Model medians were 30.75 g (ChatGPT), 30.40 g (Gemini), 29.75 g (DeepSeek) and 29.25 g (Claude). ChatGPT showed the smallest bias, the narrowest limits of agreement, and the highest correlation with clinicians' calculation. Only ChatGPT met the predefined ±5% equivalence criterion, whereas Gemini and DeepSeek achieved equivalence within a ±10% margin. Claude displayed the largest negative bias and the widest dispersion.
Conclusions
ChatGPT most accurately approximated clinicians' carbohydrate calculation among the tested AI models and fulfilled strict clinical equivalence criteria. Although the other models tended to underestimate carbohydrate content, their mean deviations remained within clinically acceptable limits. These findings suggest that AI tools, particularly ChatGPT, may serve as useful adjuncts for carbohydrate counting for people with type 1 diabetes, supporting self‐management.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Substanzen
Autor:innen und Institutionen
- Unit of Pediatrics, San Salvatore Hospital, L'Aquila, Italy.
- Department of Biotechnological and Applied Clinical Sciences, University of L'Aquila, L'Aquila, Italy.
- Dietetics Service, San Salvatore Hospital, L'Aquila, Italy.
- Diabetes Unit, San Salvatore Hospital, L'Aquila, Italy.
- Department of Life, Health and Environmental Sciences, University of L'Aquila, L'Aquila, Italy.
- IRCCS Neuromed, Pozzilli (IS), Italy.
ORCID-Profile: Luca Zagaroli, Nicholas Caione, Federica Guerra, Marco Giorgio Baroni, Maria Laura Iezzi, Maurizio Delvecchio
Zitieren
APA
Zagaroli, L., Caione, N., Zara, S., Guerra, F., Zugaro, A., Baroni, M. G., Iezzi, M. L., & Delvecchio, M. (2026). Accuracy of ChatGPT , Gemini, Claude and DeepSeek in Carbohydrate Counting. Diabetes, Obesity and Metabolism. https://doi.org/10.1111/dom.70747
Vancouver
Zagaroli L, Caione N, Zara S, Guerra F, Zugaro A, Baroni MG, et al. Accuracy of ChatGPT , Gemini, Claude and DeepSeek in Carbohydrate Counting. Diabetes, Obesity and Metabolism. 2026. doi:10.1111/dom.70747
BibTeX
@article{Zagaroli2026,
author = {Zagaroli, Luca and Caione, Nicholas and Zara, Sabina and Guerra, Federica and Zugaro, Antonella and Baroni, Marco Giorgio and Iezzi, Maria Laura and Delvecchio, Maurizio},
title = {Accuracy of ChatGPT , Gemini, Claude and DeepSeek in Carbohydrate Counting},
journal = {Diabetes, Obesity and Metabolism},
year = {2026},
doi = {10.1111/dom.70747}
}
EUVIMED