AI-based digital image dietary assessment methods compared to humans and ground truth: a systematic review.

OBJECTIVE: Human error estimating food intake is a major source of bias in nutrition research. Artificial intelligence (AI) methods may reduce bias, but the overall accuracy of AI estimates is unknown. This study was a systematic review of peer-reviewed journal articles comparing fully automated AI-based (e.g. deep learning) methods of dietary assessment from digital images to human assessors and ground truth (e.g. doubly labelled water). MATERIALS AND METHODS: Literature was searched through May 2023 in four electronic databases plus reference mining. Eligible articles reported AI estimated volume, energy, or nutrients. Independent investigators screened articles and extracted data. Potential sources of bias were documented in absence of an applicable risk of bias assessment tool. RESULTS: Database and hand searches identified 14,059 unique publications; fifty-two papers (studies) published from 2010 to 2023 were retained. For food detection and classification, 79% of papers used a convolutional neural network. Common ground truth sources were calculation using nutrient tables (51%) and weighed food (27%). Included papers varied widely in food image databases and results reported, so meta-analytic synthesis could not be conducted. Relative errors were extracted or calculated from 69% of papers. Average overall relative errors (AI vs. ground truth) ranged from 0.10% to 38.3% for calories and 0.09% to 33% for volume, suggesting similar performance. Ranges of relative error were lower when images had single/simple foods. CONCLUSIONS: Relative errors for volume and calorie estimations suggest that AI methods align with - and have the potential to exceed - accuracy of human estimations. However, variability in food image databases and results reported prevented meta-analytic synthesis. The field can advance by testing AI architectures on a limited number of large-scale food image and nutrition databases that the field determines to be adequate for training and testing and by reporting accuracy of at least absolute and relative error for volume or calorie estimations.

These results suggest that AI methods are in line with ­ and have the potential to exceed ­ accuracy of human estimations of nutrient content based on digital food images.Variability in food image databases used and results reported prevented meta-analytic synthesis.The field can advance by testing AI architectures on a limited number of large-scale food image and nutrition databases that the field determines to be accurate and by reporting accuracy of at least absolute and relative error for volume or calorie estimations.Overall, the tools currently available need more development before deployment as stand-alone dietary assessment methods in nutrition research or clinical practice.

Vitamin D intakes and health outcomes in infants and preschool children: Summary of an evidence report.

BACKGROUND: A systematic review was commissioned to support an international expert group charged to update the Food and Agriculture Organisation of the United Nations (FAO)/World Health Organisation (WHO)'s vitamin D intake recommendations for children aged 0-4 years. MATERIALS AND METHODS: Multiple electronic databases were searched to capture studies published from database inception to the 2nd week of June 2020 according to key questions formulated by the FAO/WHO. Relevant studies were summarised and synthesised by key questions and by health outcomes using the Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) approach. RESULTS: The 146 included studies examined the effects of different vitamin D intake levels on a variety of health outcomes (e.g. infectious disease, growth, neurodevelopment, rickets, and bone mineral density), and on outcomes for setting vitamin D upper limits (e.g. hypercalcemia, hypercalciuria, and nephrocalcinosis). For most outcomes, the strength of evidence was low or very low. Evidence was rated moderate for the effect of daily vitamin D supplementation on raising serum 25(OH)D concentrations, and a random-effects meta-regression analysis of 28 randomised controlled trials (mostly in infants 0-12 months) showed that each 100 IU/d increase in vitamin D supplementation was associated with an average of 1.92 (95% CI 0.28, 3.56) nmol/L increase in achieved 25-hydroxy-vitaminn D (25[OH]D) concentration (n = 53 intervention arms; p = .022) with large residual heterogeneity (I2 = 99.39%). Evidence was very low on two of the upper limit outcomes - hypercalcemia and hypercalciuria. CONCLUSIONS: The evidence report provided the expert group with a foundation and core set of data to begin their work to set vitamin D nutrient reference values. To move the field forward, future studies should use standardised 25(OH)D assay measurements and should examine the relationship between long-term vitamin D status and health outcomes.Key MessagesResults of a large complex systematic review suggest the current totality of evidence from trials and prospective observational studies do not reach sufficient certainty level to support a causal relationship between vitamin D intake and asthma, wheeze, eczema, infectious diseases, or rickets (most trials reported no rickets) in generally healthy infants and young children.In this systematic review, the only body of evidence that reached a moderate level of certainty was regarding the effect of daily vitamin D supplementation (vitamin D3 or D2 supplements to infants/children) on increasing serum 25(OH)D concentrations. However, currently there is no consensus on the definitions of vitamin D status, e.g. deficiency, insufficiency, sufficiency and toxicity, based on serum 25(OH)D concentrations.This systematic review provided an international expert group a foundation and core set of data through intake-response modelling to help set vitamin D nutrient reference values for infants and children up to 4 years of age.

Calcium Intake and Metabolism in Infants and Young Children: A Systematic Review of Balance Studies for Supporting the Development of Calcium Requirements.

Determining calcium requirements for infants and children is vital due to high calcium needs for growth. Balance studies enable comprehensive measurement of calcium metabolism and can support nutrient requirement development. This systematic review summarizes evidence from mass balance and isotopic studies in children aged 0-4 y to address key questions on calcium loss and absorption/retention identified by an expert group developing calcium requirements. Literature searches were implemented in multiple electronic databases to June 2020. Balance studies assessing calcium intake, loss, absorption, or retention in healthy children were eligible. A newly developed risk-of-bias assessment tool was used for balance studies, and a modified Grades of Recommendation, Assessment, Development, and Evaluation approach determined strength of evidence. Altogether, 23 studies (15 mass balance; 8 isotope) with 485 total participants were included. Only 3 studies were of children >6 mo. Mass balance studies suggested infant feed components may influence calcium balance. The random-effects model meta-regression on 42 mass balance study arms showed an average net calcium retention of 40.4% among infants aged 0-6 mo (ß = 0.404 [95% CI: 0.302, 0.506]). Isotope studies suggested calcium intake of 240 to 400 mg/d may promote optimal calcium absorption with minimal loss, and intake from human milk may lead to greater absorption and retention efficacy than formula or solid foods. Most studies had low risk of bias. Strength of evidence was low due to variability in infant feedings, limited endogenous and dermal calcium loss measures, and few studies isolating calcium effects. To improve certainty of the body of evidence, more balance studies isolating effects of calcium intake in this age group are needed. Future work on calcium needs should incorporate both balance measures and biological endpoints of importance (e.g. bone mineral density or content) to determine adequate calcium intake for growth in infants and children.

Effects of 100% Orange Juice on Markers of Inflammation and Oxidation in Healthy and At-Risk Adult Populations: A Scoping Review, Systematic Review, and Meta-analysis.

One hundred percent orange juice (OJ) has no added sugar, naturally contains flavonoids and ascorbic acid, and can modulate the body's oxidative and inflammatory systems. This scoping review, systematic review, and meta-analysis investigated associations between 100% OJ and markers of inflammation or oxidation in healthy adults and those at risk for chronic diseases. The study followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement and scoping review extension. Literature in English was searched to July 2021 in Embase and 4 Ovid platform databases. Clinical and observational studies of any duration were eligible. Cochrane Collaboration tools were used to assess the risk of bias in controlled trials. Strength of evidence was determined using the Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) approach. The scoping review presents a qualitative synthesis of evidence in summary and results tables. Twenty-one interventional studies (16 controlled trials and 5 before-after studies) conducted in 307 healthy and 327 at-risk participants were included. Six common markers [C-reactive protein (CRP) or high-sensitivity CRP (hs-CRP), IL-6, TNF-α, malondialdehyde (MDA), oxidized LDL (oxLDL), and antioxidant capacity] measured across 16 studies were systematically reviewed, and results were synthesized narratively. Random-effects model meta-analyses were conducted on 10 studies reporting hs-CRP, IL-6, and/or MDA. After consuming 100% OJ, healthy and at-risk participants showed significantly lower IL-6 concentrations (pooled net difference: -1.51 pg/mL; 95% CI: -2.31, -0.70) and lower, but nonsignificant, hs-CRP (pooled net change: -0.58 mg/L; 95% CI: -1.22, 0.05) and MDA (crossover trials pooled net difference: -0.06 µmol/L; 95% CI: -0.19, 0.08). Findings suggest that 100% OJ may reduce inflammation, but results should be interpreted with caution due to moderate risk of bias, very low strength of evidence, and the low number of subjects. This study was registered on PROSPERO (https://www.crd.york.ac.uk/prospero/) as CRD42021235438.

Effectiveness of multimodal nutrition interventions during pregnancy to achieve 2009 Institute of Medicine gestational weight gain guidelines: a systematic review and meta-analysis.

BACKGROUND: In 2009, the Institute of Medicine (IOM) published a revision to its 1990 recommendations on gestational weight gain (GWG). The objective of this review is to update a previous systematic review and meta-analysis to evaluate the effectiveness of nutrition interventions in achieving recommended GWG. METHODS: We conducted updated literature searches in MEDLINE® (2012 through 2019), Web of Science (2012 to 6 February 2017), Embase (2016 through 2019), and Cochrane Central Register of Controlled Trials (2012 through 2019). Literature published before January 2012 was identified from a published systematic review. We included controlled trials conducted in the U.S. or Canada among generally healthy pregnant women that compared nutrition interventions with or without exercise to controls (e.g., usual care) and reported total GWG or rate of GWG based on the 2009 IOM GWG guidelines. Two independent investigators conducted screening, data extraction, and risk-of-bias (ROB) assessment. Random-effects meta-analyses were conducted when data were sufficient. RESULTS: Eighteen unique studies were included, of which 11 were conducted in women with overweight or obesity. Nutrition interventions, compared to controls, had a similar effect on total GWG (mean difference = -1.24 kg; 95% CI [-2.65, 0.18]; I2=67.6%) but significantly decreased second and third trimester rate of GWG (-0.07 kg/week; 95% CI [-0.12, -0.03]; I2=54.7%). Nutrition interventions also reduced the risk of exceeding IOM's rate of GWG targets (pooled RR = 0.71; 95% CI [0.55, 0.92]; I2=86.3%). Meta-analyses showed no significant differences in achieving IOM's total GWG or any secondary outcome (e.g., preterm birth or small/large for gestational age) between groups. Most studies were assessed as having some or high ROB in at least two domains. CONCLUSION: Multimodal nutrition interventions designed to meet the 2009 IOM's GWG targets may decrease the rate of GWG over the second and third trimesters but may not decrease total GWG.Key messagesExcessive gestational weight gain is associated with higher risk of many adverse maternal and fetal outcomes and represents a public health concern in the United States and Canada.Nutrition interventions designed to meet the 2009 IOM GWG guidelines may decrease the rates of GWG over the second and third trimesters but may not be effective at reducing total GWG.

Safety of Using Enteral Nutrition Formulations Containing Dietary Fiber in Hospitalized Critical Care Patients: A Systematic Review and Meta-Analysis.

Enteral nutrition (EN) is the preferred route of nutrition support for patients with critical illness undergoing intensive care. Experts in the field caution against using fiber during EN because of perceived adverse patient outcomes; however, a comprehensive assessment of this topic is not evident to date. In this systematic review and meta-analysis, we searched four databases from inception to April 20, 2020, for studies on adverse events or health outcomes associated with using EN formulations containing fiber in hospitalized adults with critical illness. Nineteen articles were included. Random-effects meta-analysis models showed significantly lower diarrhea scores for fiber groups compared with nonfiber groups (pooled mean difference: -2.78; 95% CI, -4.10 to -1.47) but mixed results for risk of diarrhea between groups, depending on measures used for diarrhea (Hart and Dobb scale, pooled risk ratio [RR]: 0.68; 95% CI, 0.45-1.02; other diarrhea scales, pooled RR: 0.42; 95% CI, 0.20-0.89). Models showed 39% lower risk of gastrointestinal (GI) complications overall for fiber compared with nonfiber groups (pooled RR: 0.61; 95% CI, 0.47-0.79) but no group differences for individual GI complications, mortality, and intensive care unit or hospital length of stay. Analyses stratified by soluble- or mixed-fiber interventions reduced heterogeneity in models but showed identical conclusions. EN formulas with fiber may help reduce incidence and severity of diarrhea and GI complications overall in critically ill patients, without increased risk of other adverse events. Bias among specific GI measures indicates more high-quality studies are needed to verify these conclusions.

Vitamin D and Calcium in Children 0-36 Months: A Scoping Review of Health Outcomes.

METHODS: Updated literature searches were conducted across 5 electronic databases to identify all randomized controlled trials, cohort studies, nested case-control or case-cohort studies, and systematic reviews published after the 2009 U.S. Agency for Healthcare Research and Quality's (AHRQ) evidence report, Vitamin D and Calcium: a Systematic Review of Health Outcomes. In total, 65 studies were summarized and analyzed in the present review. RESULTS: There was a steady increasing trend in the number of publications reporting outcomes related to skeletal health, growth, and infectious disease from 2008 to 2019, although the number of published vitamin D studies was much larger than calcium studies. CONCLUSIONS: The data presented in this review can facilitate assessment of the variety of outcomes and the amount of potentially useful literature available for each outcome. The results of this scoping review can be used to guide the undertaking of the subsequent systematic reviews.