Health AI Central
Share:
Peer-Reviewed Publication
J Diabetes Sci Technol2026;19322968261462556.July 6, 2026Journal Article

Using a Fine-Tuned Commercial Artificial Intelligence Model to Assess Nutrients from Photographs of Japanese Meals.

Daniel R Lane1, Yuexiang Ji1, Naoto Otaki2,3, Maiko Kitagawa4, Kenji Obayashi5, Keigo Saeki5, Toshimasa Yamauchi6, Masaomi Nangaku7, Kayo Waki1,8
1Graduate School of Medicine, The University of Tokyo, Bunkyo-ku, Japan.
2Department of Food Sciences and Nutrition, School of Food Sciences and Nutrition, Mukogawa Women's University, Hyogo, Japan.
3Research Institute for Health Science, Mukogawa Women's University, Hyogo, Japan.
4Oura Internal Medicine Clinic, Nara, Japan.
5Department of Epidemiology, Nara Medical University School of Medicine, Japan.
6Department of Diabetes and Metabolic Diseases, Graduate School of Medicine, The University of Tokyo, Bunkyo-ku, Japan.
7Division of Nephrology and Endocrinology, Graduate School of Medicine, The University of Tokyo, Bunkyo-ku, Japan.
8Graduate School of Medicine, The University of Tokyo Hospital, Bunkyo-ku, Japan.
PubMed PMC DOI PDF

Abstract

To reduce meal logging burden in diet interventions, we fine-tuned OpenAI's GPT-4o on 1269 Japanese meal photographs (train/val/eval: 912/252/105) to estimate nutrients, using weighed food records or dietitian estimates as ground truth, and compared it with 27 non-fine-tuned models and a human dietitian. Non-fine-tuned models did poorly for fiber. Most models did well for carbohydrates, protein, a…

Create a free account to keep reading

Free members get 10 full research views every month across publications, clinical trials, FDA clearances, adverse events, and NIH grants. No credit card required.

Create Free AccountSign In

Want unlimited research access? See Pro plans

Data Accuracy Notice: Research intelligence on Health AI Central is aggregated from public sources (PubMed, ClinicalTrials.gov, FDA, NIH, CMS, and others) and refreshed nightly. Classifications and derived metrics are produced by automated methods described in our Methodology. We recommend verifying critical data points against the primary sources before making decisions.