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 > Artículos y libros
Universidad Internacional Iberoamericana México > Investigación > Producción Científica
Universidad Internacional Iberoamericana Puerto Rico > Investigación > Producción Científica
Universidad Internacional do Cuanza > Investigación > Producción Científica
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.uneatlantico.es/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 > Artículos y libros Universidad Internacional Iberoamericana México > Investigación > Producción Científica Universidad Internacional Iberoamericana Puerto Rico > Investigación > Producción Científica Universidad Internacional do Cuanza > Investigación > Producción Científica Universidad de La Romana > Investigación > Producción Científica |
| Depositado: | 03 Dic 2025 23:30 |
| Ultima Modificación: | 03 Dic 2025 23:30 |
| URI: | https://repositorio.uneatlantico.es/id/eprint/17880 |
Acciones (logins necesarios)
 |
Ver Objeto |
<a href="/17889/1/PIIS1879406825006344.pdf" class="ep_document_link"><img class="ep_doc_icon" alt="[img]" src="/style/images/fileicons/text.png" border="0"/></a>
en
open
Introduction Cancer in older adults is often associated with functional limitations, geriatric syndromes, poor self-rated health, vulnerability, and frailty, and these conditions might worsen treatment-related side effects. Recent guidelines for patients with cancer during and after treatment have documented the beneficial effects of exercise to counteract certain side effects; however, little is known about the role of exercise during cancer treatment in older adults. Materials and Methods This is a multicentre randomised controlled trial in which 200 participants will be allocated to a control group or an intervention group (the sample size has been calculated to detect a clinical difference of 1 point in Short Physical Performance Battery (SPPB) score, assuming an α error of 0.05, a β error of 0.20, and a 10 % loss rate). Patients aged ≥70 years, diagnosed with any type of solid cancer and candidates for systemic treatment are eligible. Subjects in the intervention group are invited to participate in a 12-week supervised multicomponent exercise programme in addition to receiving usual care. Study assessments are conducted at baseline and three months. The primary outcome measure is physical function as assessed by the SPPB. Secondary outcome measures include comprehensive geriatric assessment scores (including social situation, basic and instrumental activities of daily living, cognitive function, depression, nutritional status, polypharmacy, geriatric syndromes, pain, and emotional distress), anthropometric characteristics, frailty status, physical fitness, physical activity, cognitive function, quality of life, fatigue, and nutritional status. Study assessments also include analysis of inflammatory, endocrine, and nutritional mediators in serum and plasma as potential frailty biomarkers at mRNA and protein levels and multiparametric flow cytometric analysis to measure immunosenescence markers on T and NK cells. Discussion This study seeks to extend our knowledge on exercise interventions during systemic anticancer treatment in patients over 70 years of age. Results from this research will guide the management of older adults during systemic treatment in hospitals seeking to enhance the standard of care.
Julia García-García mail , Ana Rodriguez-Larrad mail , Maren Martinez de Rituerto Zeberio mail , Jenifer Gómez Mediavilla mail , Borja López-San Vicente mail , Nuria Torrego Artola mail , Izaskun Zeberio Etxetxipia mail , Irati Garmendia mail , Ainhoa Alberro mail , David Otaegui mail , Francisco Borrego Rabasco mail , María M. Caffarel mail , Kalliopi Vrotsou mail , Jon Irazusta mail , Haritz Arrieta mail , Mireia Peláez mail mireia.pelaez@uneatlantico.es, Jon Belloso mail , Laura Basterretxea mail ,
García-García
<a href="/28567/1/3051020.pdf" class="ep_document_link"><img class="ep_doc_icon" alt="[img]" src="/style/images/fileicons/text.png" border="0"/></a>
en
open
Background Physical inactivity and suboptimal diet in pregnancy are important modifiable risk factors for gestational diabetes, a major contributor to pregnancy complications. Objectives We aimed to assess the effects of physical activity and/or diet-based lifestyle interventions during pregnancy on gestational diabetes and if these vary by maternal (body mass index, age, parity, ethnicity, education) and intervention characteristics using individual participant data meta-analysis of randomised trials, and a cost-effectiveness analysis. Data sources International Weight Management in Pregnancy Collaborative Network database was updated by searching major databases from February 2017 to March 2022. Review methods The main outcomes were gestational diabetes by any criteria and by the National Institute for Health and Care Excellence. Other outcomes were gestational diabetes as per International Association of Diabetes in Pregnancy Study Group and maternal and perinatal outcomes. We performed a two-stage random-effects individual participant data meta-analysis to obtain summary estimates (odds ratio) with 95% confidence intervals. Study quality of included trials was assessed, and heterogeneity summarised using τ2. Where possible, we added the aggregate data from non-individual participant data trials to the meta-analysis. We ranked interventions by effectiveness using network meta-analysis and undertook model-based economic evaluation to assess cost-effectiveness. The cost-effectiveness analysis took an NHS cost perspective compared an overall lifestyle intervention versus usual care with a time horizon covering the beginning of pregnancy until the discharge of the mother and infant from the hospital following delivery. Results Ninety-two trials (32,284 women) were included; 54 (23,698 women) provided individual participant data. Lifestyle interventions reduced the odds of gestational diabetes (any criteria) by 10% in individual participant data trials (odds ratio 0.90, 95% confidence interval 0.80 to 1.02, 54 studies, 23,361 women), and the findings reached statistical significance when non-individual participant data were included (odds ratio 0.81, 95% confidence interval 0.73 to 0.89, 92 studies, 31,947 women). Physical activity significantly reduced the odds of gestational diabetes by 36% (odds ratio 0.64; 95% confidence interval 0.48 to 0.84), and diet by 19% (odds ratio 0.81; 0.69 to 0.96), but not mixed interventions. Women with middle (odds ratio 0.68, 95% confidence interval 0.51 to 0.90) and high educational level (odds ratio 0.71, 95% confidence interval 0.54 to 0.93) benefited more than those with low educational status, and no differences by maternal body mass index, age, parity or ethnicity. There was no significant reduction in gestational diabetes defined by National Institute for Health and Care Excellence criteria (odds ratio 0.98, 95% confidence interval 0.84 to 1.13) in individual participant data trials. For gestational diabetes defined using International Association of Diabetes in Pregnancy Study Group criteria, interventions reduced gestational diabetes by 14% (odds ratio 0.86, 95% confidence interval 0.75 to 0.97, τ2 = 0.00, 16 studies, 6174 women) in individual participant data trials and by 17% (odds ratio 0.83, 95% confidence interval 0.72 to 0.95, τ2 = 0.01, 25 studies, 7883 women) when non-individual participant data trials were added. Overall, physical activity reduced caesarean section (odds ratio 0.83; 0.72 to 0.96), small-for-gestational age (odds ratio 0.72; 0.56 to 0.92) and large-for-gestational age babies (odds ratio 0.81; 0.71 to 0.94); diet-based interventions reduced any preterm birth (odds ratio 0.37; 0.20 to 0.68) compared to controls. No differences were observed for other outcomes. Lifestyle interventions were on average more expensive and more effective at averted gestational diabetes and major outcome averted compared to usual care. Limitations We could not identify the specific intervention components and delivery methods associated with improved outcomes, due to variations in reporting. Conclusion Lifestyle interventions in pregnancy prevent gestational diabetes, and the effects vary according to the definition of gestational diabetes. Physical activity-based interventions may be the most effective.
John Allotey mail , Dyuti Coomar mail , Joie Ensor mail , Chidubem Okeke Ogwulu mail , Gabriel Ruiz Calvo mail , Mark Monahan mail , Valencia Kabeya mail , Rachel McNeill mail , Anna Boath mail , Ghadir Mahmoud mail , Cheryce Harrison mail , Mahnaz Bahri Khomami mail , Helena Teede mail , Nicola Heslehurst mail , Graham A Hitman mail , Sharon Anne Simpson mail , Krish Nirantharakumar mail , Julie Dodds mail , Kelly C Allison mail , Garry Shen mail , Elisabetta Petrella mail , Fabio Facchinetti mail , Christina Vinter mail , Mireia Peláez mail mireia.pelaez@uneatlantico.es, Dorte Møller Jensen mail , Narges Sadat Motahari-Tabari mail , Tarja I Kinnunen mail , Jonatan R Ruiz mail , Annick Bogaerts mail , Kristina Martha Renault mail , Alka Kothari mail , Jose Guilherme Cecatti mail , Fionnuala M McAuliffe mail , Suzanne Phelan mail , Lucilla Poston mail , Ana Pilar Betrán mail , Ngawai Moss mail , Stamatina Iliodromiti mail , Frances Austin mail , Nuria García de la Torre mail , Alfonso Luis Calle Pascual mail , Javier Zamora mail , Tracy Roberts mail , Richard D Riley mail , Shakila Thangaratinam mail ,
Allotey
en
close
Objectives Common mental disorders (CMDs), including depression and anxiety, are highly prevalent in primary care, yet access to psychological therapies, which are the first-line treatment for these conditions, remains limited. This study evaluated the effectiveness of the Clinical Psychology in Primary Care Programme in Cantabria during its first year of implementation, which integrated clinical psychologists into primary care teams to deliver brief, evidence-based interventions. Design Naturalistic observational pre–post study in primary care. Methods A total of 1149 patients (66% women; M = 44 years) were assessed with the PHQ-9 and GAD-7 before and after a brief psychological intervention of up to eight sessions per patient delivered individually or in groups. Linear mixed-effects models examined pre–post changes and moderators (intervention format, age, sex). Reliable change, deterioration and recovery were calculated. Results Significant and clinically meaningful reductions were observed in depressive (ΔPHQ-9 = −7.8) and anxiety symptoms (ΔGAD-7 = −7.1; p < .001). Individual therapy produced greater improvements than group interventions, and younger participants showed slightly larger gains; sex showed no effect. Moderate variability in outcomes was observed across therapists. Among completers, large effect sizes were found (dRM ≈ 1.3); 73% achieved reliable improvement and 51% met reliable recovery criteria, while deterioration rates were below 1%. Conclusions Findings support the consolidation and expansion of the integration of clinical psychologists into primary care, providing evidence that the psychological interventions delivered in this context are a feasible, equitable and effective approach to treating CMDs in the Spanish public health system.
César González‐Blanch mail cesar.gonzalezblanch@uneatlantico.es, Noelia Rodríguez‐Pérez mail , María Camino‐Sánchez mail , Rosario Bengochea-Seco mail , Cintia Montes‐Novoa mail , David Gil-Sanz mail david.gil@uneatlantico.es, Silvia Pérez‐Monzón mail , Blanca Uriz‐Zafra mail , Mikel Muñiz‐Videchea mail , Paula Díaz‐Gómez mail , Amador Priede mail ,
González‐Blanch
en
close
Obesity is recognised to be a risk factor for breast cancer since adipose tissue influences the tumour microenvironment. This study aims to investigate the effect of the secretome of 3T3-L1 adipocytes untreated or treated with liquorice root extract (LRE), containing flavonoids, phenolic acids, and saponins on MCF-7 breast cancer cells. By treating adipocytes with LRE, the secretion of certain pro-tumorigenic factors like IGFBP-6, resistin, and VEGF was reduced. MCF-7 cells exposed to conditioned medium from LRE-treated adipocytes exhibited an increase in reactive oxygen species levels, downregulation of the Nrf2 antioxidant pathway, and increased autophagy. Those conditions reduced cell viability, migration, and colony formation. Additionally, there was downregulation of genes associated with oestrogen signalling and tumour-related processes, including CYP19A1 (aromatase), ERα, Her2, and components of the renin–angiotensin system (RAS). These findings suggest that LRE can modulate the adipocyte secretome to influence breast cancer cell behaviour under obesity-related in vitro conditions.
Danila Cianciosi mail , Yasmany Armas Diaz mail , Bei Yang mail , Zexiu Qi mail , Ge Chen mail , José L. Quiles mail jose.quiles@uneatlantico.es, Massimiliano Gasparrini mail , Manuela Cassotta mail manucassotta@gmail.com, Rubén Calderón Iglesias mail ruben.calderon@uneatlantico.es, Maurizio Battino mail maurizio.battino@uneatlantico.es, Francesca Giampieri mail francesca.giampieri@uneatlantico.es,
Cianciosi
en
close
Theaflavins are characteristic polyphenols formed during black tea fermentation. They play an important role in tea color and quality. With the increasing global consumption of black tea, interest in fermentation-derived polyphenols has grown. Recent studies have examined the chemical properties and reported bioactivities of theaflavins. However, current findings remain scattered and lack a unified food chemistry perspective. This review summarizes the formation chemistry of theaflavins during black tea fermentation. It compares the key factors affecting their composition and abundance, including catechin precursors, enzymatic oxidation, and fermentation conditions. Structural features of different theaflavin derivatives are discussed. Particular attention is given to galloylated theaflavins, which show higher reactivity in experimental systems. Reported bioactivities are briefly addressed to illustrate structure-activity relationships. Challenges related to production standardization, stability, and bioaccessibility are also highlighted.
Yuxuan Zhao mail , Jingyimei Liang mail , Hui Cao mail , Jianbo Xiao mail ,
Zhao