Tailoring the Nutritional Composition of Italian Foods to the US Nutrition5k Dataset for Food Image Recognition: Challenges and a Comparative Analysis.

BACKGROUND: Training of machine learning algorithms on dish images collected in other countries requires possible sources of systematic discrepancies, including country-specific food composition databases (FCDBs), to be tackled. The US Nutrition5k project provides for ~5000 dish images and related dish- and ingredient-level information on mass, energy, and macronutrients from the US FCDB. The aim of this study is to (1) identify challenges/solutions in linking the nutritional composition of Italian foods with food images from Nutrition5k and (2) assess potential differences in nutrient content estimated across the Italian and US FCDBs and their determinants. METHODS: After food matching, expert data curation, and handling of missing values, dish-level ingredients from Nutrition5k were integrated with the Italian-FCDB-specific nutritional composition (86 components); dish-specific nutrient content was calculated by summing the corresponding ingredient-specific nutritional values. Measures of agreement/difference were calculated between Italian- and US-FCDB-specific content of energy and macronutrients. Potential determinants of identified differences were investigated with multiple robust regression models. RESULTS: Dishes showed a median mass of 145 g and included three ingredients in median. Energy, proteins, fats, and carbohydrates showed moderate-to-strong agreement between Italian- and US-FCDB-specific content; carbohydrates showed the worst performance, with the Italian FCDB providing smaller median values (median raw difference between the Italian and US FCDBs: -2.10 g). Regression models on dishes suggested a role for mass, number of ingredients, and presence of recreated recipes, alone or jointly with differential use of raw/cooked ingredients across the two FCDBs. CONCLUSIONS: In the era of machine learning approaches for food image recognition, manual data curation in the alignment of FCDBs is worth the effort.

Are Major a Posteriori Dietary Patterns Reproducible in the Italian Population? A Systematic Review and Quantitative Assessment.

Although a posteriori dietary patterns (DPs) naturally reflect actual dietary behavior in a population, their specificity limits generalizability. Among other issues, the absence of a standardized approach to analysis have further hindered discovery of genuinely reproducible DPs across studies from the same/similar populations. A systematic review on a posteriori DPs from principal component analysis or exploratory factor analysis (EFA) across study populations from Italy provides the basis to explore assessment and drivers of DP reproducibility in a case study of epidemiological interest. First to our knowledge, we carried out a qualitative (i.e., similarity plots built on text descriptions) and quantitative (i.e., congruence coefficients, CCs) assessment of DP reproducibility. The 52 selected articles were published in 2001-2022 and represented dietary habits in 1965-2022 from 70% of the Italian regions; children/adolescents, pregnancy/breastfeeding women, and elderly were considered in 15 articles. The included studies mainly derived EFA-based DPs on food groups from food frequency questionnaires and were of "good quality" according to standard scales. Based on text descriptions, the 186 identified DPs were collapsed into 113 (69 food-based and 44 nutrient-based) apparently different DPs (39.3% reduction), later summarized along with the 3 "Mixed-Salad/Vegetable-based Patterns," "Pasta-and-Meat-oriented/Starchy Patterns," and "Dairy Products" and "Sweets/Animal-based Patterns" groups, by matching similar food-based and nutrient-based groups of collapsed DPs. Based on CCs (215 CCs, 68 DPs, 18 articles using the same input lists), all pairs of DPs showing the same/similar names were at least "fairly similar" and ∼81% were "equivalent." The 30 "equivalent" DPs ended up into 6 genuinely different DPs (80% reduction) that targeted fruits and (raw) vegetables, pasta and meat combined, and cheese and deli meats. Such reduction reflects the same study design, list of input variables, and DP identification method followed across articles from the same groups. This review was registered at PROSPERO as CRD42022341037.

Does Pizza Consumption Favor an Improved Disease Activity in Rheumatoid Arthritis?

To our knowledge, no studies so far have investigated the role of pizza and its ingredients in modulating disease activity in rheumatoid arthritis (RA). We assessed this question via a recent cross-sectional study including 365 participants from Italy, the birthplace of pizza. Multiple robust linear and logistic regression models were fitted with the tertile consumption categories of each available pizza-related food item/group (i.e., pizza, refined grains, mozzarella cheese, and olive oil) as independent variables, and each available RA activity measure (i.e., the Disease Activity Score on 28 joints with C-reactive protein (DAS28-CRP), and the Simplified Disease Activity Index (SDAI)) as the dependent variable. Stratified analyses were carried out according to the disease severity or duration. Participants eating half a pizza >1 time/week (vs. ≤2 times/month) reported beneficial effects on disease activity, with the significant reductions of ~70% (overall analysis), and 80% (the more severe stratum), and the significant beta coefficients of -0.70 for the DAS28-CRP, and -3.6 for the SDAI (overall analysis) and of -1.10 and -5.30 (in long-standing and more severe RA, respectively). Among the pizza-related food items/groups, mozzarella cheese and olive oil showed beneficial effects, especially in the more severe stratum. Future cohort studies are needed to confirm this beneficial effect of pizza and related food items/groups on RA disease activity.