Image-Based Dietary Energy and Macronutrients Estimation with ChatGPT-5: Cross-Source Evaluation Across Escalating Context Scenarios
Artículo
Materias > Ingeniería
Universidad Europea del Atlántico > Investigación > Producción Científica
Universidad Internacional Iberoamericana México > Investigación > Producción Científica
Universidad Internacional Iberoamericana Puerto Rico > Investigación > Producción Científica
Universidad Internacional do Cuanza > Investigación > Artículos y libros
Universidad de La Romana > Investigación > Producción Científica
Abierto
Inglés
Background/Objectives: Estimating energy and macronutrients from food images is clinically relevant yet challenging, and rigorous evaluation requires transparent accuracy metrics with uncertainty and clear acknowledgement of reference data limitations across heterogeneous sources. This study assessed ChatGPT-5, a general-purpose vision-language model, across four scenarios differing in the amount and type of contextual information provided, using a composite dataset to quantify accuracy for calories and macronutrients. Methods: A total of 195 dishes were evaluated, sourced from Allrecipes.com, the SNAPMe dataset, and Home-prepared, weighed meals. Each dish was evaluated under Case 1 (image only), Case 2 (image plus standardized non-visual descriptors), Case 3 (image plus ingredient lists with amounts), and Case 4 (replicates Case 3 but excluding the image). The primary endpoint was kcal Mean Absolute Error (MAE); secondary endpoints included Median Absolute Error (MedAE) and Root Mean Square Error (RMSE) for kcal and macronutrients (protein, carbohydrates, and lipids), all reported with 95% Confidence Intervals (CIs) via dish-level bootstrap resampling and accompanied by absolute differences (Δ) between scenarios. Inference settings were standardized to support reproducibility and variance estimation. Source stratified analyses and quartile summaries were conducted to examine heterogeneity by curation level and nutrient ranges, with additional robustness checks for error complexity relationships. Results and Discussion: Accuracy improved from Case 1 to Case 2 and further in Case 3 for energy and all macronutrients when summarized by MAE, MedAE, and RMSE with 95% CIs, with absolute reductions (Δ) indicating material gains as contextual information increased. In contrast to Case 3, estimation accuracy declined in Case 4, underscoring the contribution of visual cues. Gains were largest in the Home-prepared dietitian-weighed subset and smaller yet consistent for Allrecipes.com and SNAPMe, reflecting differences in reference curation and measurement fidelity across sources. Scenario-level trends were concordant across sources, and stratified and quartile analyses showed coherent patterns of decreasing absolute errors with the provision of structured non-visual information and detailed ingredient data. Conclusions: ChatGPT-5 can deliver practically useful calorie and macronutrient estimates from food images, particularly when augmented with standardized nonvisual descriptors and detailed ingredients, as evidenced by reductions in MAE, MedAE, and RMSE with 95% CIs across scenarios. The decline in accuracy observed when the image was omitted, despite providing detailed ingredient information, indicates that visual cues contribute meaningfully to estimation performance and that improvements are not solely attributable to arithmetic from ingredient lists. Finally, to promote generalizability, it is recommended that future studies include repeated evaluations across diverse datasets, ensure public availability of prompts and outputs, and incorporate systematic comparisons with non-artificial-intelligence baselines.
metadata
Rodríguez- Jiménez, Marcela; Martín-del-Campo-Becerra, Gustavo Daniel; Sumalla Cano, Sandra; Crespo-Álvarez, Jorge y Elío Pascual, Iñaki
mail
SIN ESPECIFICAR, SIN ESPECIFICAR, sandra.sumalla@uneatlantico.es, jorge.crespo@uneatlantico.es, inaki.elio@uneatlantico.es
(2025)
Image-Based Dietary Energy and Macronutrients Estimation with ChatGPT-5: Cross-Source Evaluation Across Escalating Context Scenarios.
Nutrients, 17 (22).
p. 3613.
ISSN 2072-6643
![[img]](http://repositorio.unic.co.ao/style/images/fileicons/text.png) |
Texto
nutrients-17-03613.pdf Available under License Creative Commons Attribution. Descargar (7MB) |
Resumen
Background/Objectives: Estimating energy and macronutrients from food images is clinically relevant yet challenging, and rigorous evaluation requires transparent accuracy metrics with uncertainty and clear acknowledgement of reference data limitations across heterogeneous sources. This study assessed ChatGPT-5, a general-purpose vision-language model, across four scenarios differing in the amount and type of contextual information provided, using a composite dataset to quantify accuracy for calories and macronutrients. Methods: A total of 195 dishes were evaluated, sourced from Allrecipes.com, the SNAPMe dataset, and Home-prepared, weighed meals. Each dish was evaluated under Case 1 (image only), Case 2 (image plus standardized non-visual descriptors), Case 3 (image plus ingredient lists with amounts), and Case 4 (replicates Case 3 but excluding the image). The primary endpoint was kcal Mean Absolute Error (MAE); secondary endpoints included Median Absolute Error (MedAE) and Root Mean Square Error (RMSE) for kcal and macronutrients (protein, carbohydrates, and lipids), all reported with 95% Confidence Intervals (CIs) via dish-level bootstrap resampling and accompanied by absolute differences (Δ) between scenarios. Inference settings were standardized to support reproducibility and variance estimation. Source stratified analyses and quartile summaries were conducted to examine heterogeneity by curation level and nutrient ranges, with additional robustness checks for error complexity relationships. Results and Discussion: Accuracy improved from Case 1 to Case 2 and further in Case 3 for energy and all macronutrients when summarized by MAE, MedAE, and RMSE with 95% CIs, with absolute reductions (Δ) indicating material gains as contextual information increased. In contrast to Case 3, estimation accuracy declined in Case 4, underscoring the contribution of visual cues. Gains were largest in the Home-prepared dietitian-weighed subset and smaller yet consistent for Allrecipes.com and SNAPMe, reflecting differences in reference curation and measurement fidelity across sources. Scenario-level trends were concordant across sources, and stratified and quartile analyses showed coherent patterns of decreasing absolute errors with the provision of structured non-visual information and detailed ingredient data. Conclusions: ChatGPT-5 can deliver practically useful calorie and macronutrient estimates from food images, particularly when augmented with standardized nonvisual descriptors and detailed ingredients, as evidenced by reductions in MAE, MedAE, and RMSE with 95% CIs across scenarios. The decline in accuracy observed when the image was omitted, despite providing detailed ingredient information, indicates that visual cues contribute meaningfully to estimation performance and that improvements are not solely attributable to arithmetic from ingredient lists. Finally, to promote generalizability, it is recommended that future studies include repeated evaluations across diverse datasets, ensure public availability of prompts and outputs, and incorporate systematic comparisons with non-artificial-intelligence baselines.
| Tipo de Documento: | Artículo |
|---|---|
| Palabras Clave: | calorie and macronutrient estimation; image-based dietary assessment; validation metrics (MAE, MedAE, RMSE); vision-language models |
| Clasificación temática: | Materias > Ingeniería |
| Divisiones: | Universidad Europea del Atlántico > Investigación > Producción Científica Universidad Internacional Iberoamericana México > Investigación > Producción Científica Universidad Internacional Iberoamericana Puerto Rico > Investigación > Producción Científica Universidad Internacional do Cuanza > Investigación > Artículos y libros Universidad de La Romana > Investigación > Producción Científica |
| Depositado: | 03 Dic 2025 23:32 |
| Ultima Modificación: | 03 Dic 2025 23:32 |
| URI: | https://repositorio.unic.co.ao/id/eprint/17880 |
Acciones (logins necesarios)
 |
Ver Objeto |
<a class="ep_document_link" href="/28319/1/s41598-026-45575-1_reference.pdf"><img class="ep_doc_icon" alt="[img]" src="/style/images/fileicons/text.png" border="0"/></a>
en
open
A novel approach for disease and pests detection in potato production system based on deep learning
Vulnerability of potato crops to diseases and pest infestation can affect its quality and lead to significant yield losses. Timely detection of such diseases can help take effective decisions. For this purpose, a deep learning-based object detection framework is designed in this study to identify and classify major potato diseases and pests under real-world field conditions. A total of 2,688 field images were collected from two research farms in Punjab, Pakistan, across multiple growth stages in various seasonal conditions. Excluding 285 symptoms-free images from the earliest collection led to 2,403 images which were annotated into four biotic-stress classes: blight disease (n = 630), leaf spot disease (n = 370), leafroll virus (viral symptom complex; n = 888), and Colorado potato beetle (larvae/adults; n = 515), indicating class imbalance. Several state-of-the-art models were used including YOLOv8 variants (n/s/m), YOLOv7, YOLOv5, and Faster R-CNN, and the results are discussed in relation to recent potato disease classification studies involving cropped leaf images. Stratified splitting (70% training, 20% validation, 10% testing) was applied to preserve class distribution across all subsets. YOLOv8-medium achieve the best performance with mean average precision (mAP)@0.5 of 98% on the held-out test images. Results for stable 5-fold cross-validation show a mean mAP@0.5 of 97.8%, which offers a balance between accuracy and inference time. Model robustness was evaluated using 5-fold cross-validation and repeated training with different random seeds, showing a low variance of ±0.4% mAP. Results demonstrate promising outcomes under the real-world field conditions, while, broader cross-region and cross-season validation is intended for the future.
Ahmed Abbas mail , Saif Ur Rehman mail , Khalid Mahmood mail , Santos Gracia Villar mail santos.gracia@uneatlantico.es, Luis Alonso Dzul López mail luis.dzul@uneatlantico.es, Aseel Smerat mail , Imran Ashraf mail ,
Abbas
<a href="/28323/1/s40520-026-03363-x_reference.pdf" class="ep_document_link"><img class="ep_doc_icon" alt="[img]" src="/style/images/fileicons/text.png" border="0"/></a>
en
open
Fish consumption and brain structure: a comprehensive systematic review of observational studies
Background Age-related structural changes in the human brain, including cortical atrophy, reductions in grey and white matter volumes, and the accumulation of small vessel–related lesions such as white matter hyperintensities (WMH) and cerebral microbleeds, represent critical biological substrates underlying cognitive decline and dementia. Fish consumption has been associated with slower cognitive decline and reduced risk of dementia, but a comprehensive evaluation of its relation with brain structures is lacking. Aims The aim of this study was to systematically review current scientific literature providing evidence of relation between fish intake and brain structures in human studies. Methods Studies indexed in two major electronic databases have been screened based on a combination of keywords and MeSH terms. Studies were eligible whether they assessed fish consumption in relation to brain structures in the adult populations. Results A total of 24 studies conducted predominantly on older adults met inclusion criteria. Most brain volume measures were obtained via magnetic resonance imaging (MRI) procedures. Higher fish consumption was associated with reduced severity of white matter hyperintensities (a biomarker of cerebral small vessel disease and white matter damage) and cerebral micro-bleed, preservation of certain brain areas volumes (i.e., hippocampus, temporal lobe and periventricle white matter) and cortical thickness of specific areas (i.e., precuneus, parietal, and cingulate grey matter), among others, compared to lower intake. Some analyses found no association and isolated findings suggested possible adverse associations that were not consistently replicated. Studies reporting null findings may underline the possible relevance of the overall diet (i.e., adherence to the Mediterranean diet). Conclusions Inclusion of fish in a healthy and balanced diet is associated with better white matter grades on MRI and slower progression of white matter hyperintensities and reduction of vascular-related lesions of the aging brain, suggesting a potential role in preventing neurocognitive deterioration. Heterogeneity across studies underscores the need for additional studies.
Justyna Godos mail , Giuseppe Caruso mail , Agnieszka Micek mail , Alberto Dolci mail , Zoltan Ungvari mail , Andrea Lehoczki mail , Lisandra León Brizuela mail , Evelyn Frias-Toral mail , Andrea Di Mauro mail , Mario Siervo mail , Michelino Di Rosa mail , Giuseppe Grosso mail ,
Godos
<a class="ep_document_link" href="/28563/1/fnut-13-1809163.pdf"><img class="ep_doc_icon" alt="[img]" src="/style/images/fileicons/text.png" border="0"/></a>
en
open
Background: Mild Cognitive Impairment (MCI) is viewed as a transitional stage between normal brain aging and dementia and is characterized by subtle cognitive deficits without significant impairment in daily functioning. Growing evidence supports the contribution of neuroinflammation and modifiable lifestyle factors, including diet, in the progression of cognitive decline.Aim: This study aimed to investigate the association between adherence to the Mediterranean diet, neuroinflammatory biomarkers, and MCI status in older adults.Design: Ninety-two participants were enrolled in this cross-sectional study, including 37 subjects with MCI. Dietary intake was assessed using a validated food frequency questionnaire (FFQ) and adherence to the Mediterranean diet explored through the MedDietScore. Plasma levels of TGF-β1 and TNF-α were measured by ELISA. Cognitive status was evaluated using the Mini Mental Examination (MMSE) and the Montreal Cognitive Assessment (MoCA), both adjusted for age and education. Statistical analyses included non-parametric tests, correlation analysis, and logistic regression models.Results: MCI patients showed significantly reduced plasma levels of TGF-β1 and increased TNF-α concentrations compared to other participants. After adjustment for potential confounding factors, greater adherence to the Mediterranean diet was associated with a lower likelihood of MCI in a dose–response manner (highest versus lowest adherence quartile, odds ratio: 0.07, 95% confidence interval: 0.01–0.60). Additional adjustment for inflammatory biomarkers attenuated the associations, suggesting a potential mediating role.Conclusion: Our findings showed that higher adherence to Mediterranean diet is associated with lower likelihood of being MCI. Such a relation might be, at least in part, mediated by inflammatory biomarkers. Overall, these results support the role of dietary modulation in preventive strategies against cognitive decline and progression into MCI.
Margherita Grasso mail , Francesca L’Episcopo mail , Annamaria Fidilio mail , Marco Antonio Olvera-Moreira mail , Giuseppe Toscano mail , Stefano Muratore mail , Margherita Drago mail , Sabrina Musso mail , Veronica Bentivegna mail , Lucrezia Costanzo mail , Melannie Toral-Noristz mail , Raynier Zambrano-Villacres mail , Lisandra León Brizuela mail , Giuseppe Lanza mail , Raffaele Ferri mail , Filippo Caraci mail ,
Grasso
<a href="/27825/1/s41598-026-39196-x_reference.pdf" class="ep_document_link"><img class="ep_doc_icon" alt="[img]" src="/style/images/fileicons/text.png" border="0"/></a>
en
open
Histopathological evaluation is necessary for the diagnosis and grading of prostate cancer, which is still one of the most common cancers in men globally. Traditional evaluation is time-consuming, prone to inter-observer variability, and challenging to scale. The clinical usefulness of current AI systems is limited by the need for comprehensive pixel-level annotations. The objective of this research is to develop and evaluate a large-scale benchmarking study on a weakly supervised deep learning framework that minimizes the need for annotation and ensures interpretability for automated prostate cancer diagnosis and International Society of Urological Pathology (ISUP) grading using whole slide images (WSIs). This study rigorously tested six cutting-edge multiple instance learning (MIL) architectures (CLAM-MB, CLAM-SB, ILRA-MIL, AC-MIL, AMD-MIL, WiKG-MIL), three feature encoders (ResNet50, CTransPath, UNI2), and four patch extraction techniques (varying sizes and overlap) using the PANDA dataset (10,616 WSIs), yielding 72 experimental configurations. The methodology used distributed cloud computing to process over 31 million tissue patches, implementing advanced attention mechanisms to ensure clinical interpretability through Grad-CAM visualizations. The optimum configuration (UNI2 encoder with ILRA-MIL, 256 256 patches, 50% overlap) achieved 78.75% accuracy and 90.12% quadratic weighted kappa (QWK), outperforming traditional methods and approaching expert pathologist-level diagnostic capability. Overlapping smaller patches offered the best balance of spatial resolution and contextual information, while domain-specific foundation models performed noticeably better than generic encoders. This work is the first large-scale, comprehensive comparison of weekly supervised MIL methods for prostate cancer diagnosis and grading. The proposed approach has excellent clinical diagnostic performance, scalability, practical feasibility through cloud computing, and interpretability using visualization tools.
Naveed Anwer Butt mail , Dilawaiz Sarwat mail , Irene Delgado Noya mail irene.delgado@uneatlantico.es, Kilian Tutusaus mail kilian.tutusaus@uneatlantico.es, Nagwan Abdel Samee mail , Imran Ashraf mail ,
Butt
<a href="/27915/1/csbj.0023.pdf" class="ep_document_link"><img class="ep_doc_icon" alt="[img]" src="/style/images/fileicons/text.png" border="0"/></a>
en
open
This systematic literature review (SLR) investigates the integration of deep learning (DL), vision-language models(VLMs), and multi-agent systems in the analysis of pathology images and automated report generation. The rapidadvancement of whole-slide imaging (WSI) technologies has posed new challenges in pathology, especially due to thescale and complexity of the data. DL techniques in general and convolutional neural networks (CNNs) and transform-ers in particular have significantly enhanced image analysis tasks including segmentation, classification, and detection.However, these models often lack generalizability to generate coherent, clinically relevant text, thus necessitating theintegration of VLMs and large language models (LLMs). This review examines the effectiveness of VLMs and LLMsin bridging the gap between visual data and clinical text, focusing on their potential for automating the generationof pathology reports. Additionally, multi-agent systems, which leverage specialized artificial intelligence (AI) agentsto collaboratively perform diagnostic tasks, are explored for their contributions to improving diagnostic accuracy andscalability. Through a synthesis of recent studies, this review highlights the successes, challenges, and future direc-tions of these AI technologies in pathology diagnostics, offering a comprehensive foundation for the development ofintegrated, AI-driven diagnostic workflows.
Usama Ali mail , Imran Shafi mail , Jamil Ahmad mail , Arlette Zárate Cáceres mail , Thania Chio Montero mail , Hafiz Muhammad Raza ur Rehman mail , Imran Ashraf mail ,
Ali