Statistical methods are usually applied in examining diet-disease associations, whereas factor analysis is commonly used for dietary pattern recognition. Recently, machine learning (ML) has been also proposed as an alternative technique in health classification. In this work, the predictive accuracy of statistical v. ML methodologies as regards the association of dietary patterns on CVD risk was tested. During 2001-2002, 3042 men and women (45 (sd 14) years) were enrolled in the ATTICA study. In 2011-2012, the 10-year CVD follow-up was performed among 2020 participants. Item Response Theory was applied to create a metric of combined 10-year cardiometabolic risk, the 'Cardiometabolic Health Score', that incorporated incidence of CVD, diabetes, hypertension and hypercholesterolaemia. Factor analysis was performed to extract dietary patterns, on the basis of either foods or nutrients consumed; linear regression analysis was used to assess their association with the cardiometabolic score. Two ML techniques (k-nearest-neighbor's algorithm and random-forests decision tree) were applied to evaluate participants' health based on dietary information. Factor analysis revealed five and three factors from foods and nutrients, respectively, explaining 54 and 65 % of the total variation in intake. Nutrient and food pattern regression models showed similar accuracy in correctly classifying an individual according to the cardiometabolic risk (R 2=9·6 % and R 2=8·3 %, respectively). ML techniques were superior compared with linear regression in correct classification of the individuals according to the Health Score (accuracy approximately 38 v. 6 %, respectively), whereas the two ML methods showed equal classification ability. Conclusively, ML methods could be a valuable tool in the field of nutritional epidemiology, leading to more accurate disease-risk evaluation.

In a population-based longitudinal cohort study, we tested the hypothesis that children growing up in a high-traffic polluted urban area (UA) in the Athens' basin have higher prevalence of allergies and sensitization when compared with those growing up in a Greek provincial rural area (RA). We recruited 478 and 342 children aged 8-10 living in the UA and the RA, respectively. Respiratory health was assessed by a parent-completed questionnaire in three phases: 1995-96 (phase 1), 1999-2000 (phase 2), 2003-04 (phase 3) and skin-prick testing to common indoor and outdoor aeroallergens was performed at phases 1 and 2. Reported asthma and eczema did not differ between the two areas, whereas reported hay fever was persistently more prevalent in the UA than in the RA (16.5%, 17.0%, 18.2% vs. 7.0%, 8.3%, 9.6%, respectively). Sensitization was more prevalent in the UA at both phases (19.0% vs. 12.1% in phase 1, 20.0% vs. 14.1% in phase 2). Residential area contributed independently to sensitization to >or=1 aeroallergens (OR: 0.29; 95% CI: 0.13-0.66; p = 0.003) and to polysensitization (OR: 0.28; 95% CI: 0.10-0.82; p = 0.020) in phase 1. These associations were independent of farming practices. No significant contributions were found in phase 2. Our results suggest that long-term exposure to urban environment is associated with a higher prevalence of hay fever but not of asthma or eczema. The negative association between rural living and the risk of atopy during childhood, which is independent of farming practices, implies that it is mainly driven by an urban living effect.

In the last few years, the need for processing large amount of data in nutrition science was dramatically arose. This created the need to apply, primarily, advanced analytical research methods that could enable researchers to handle the large amount of information. Dietary pattern analysis is a commonly used approach to enable and incorporate this phenomenon in nutrition research. This article reviews the most common dietary pattern's assessment statistical methods, evaluating at the same time the up-to-day knowledge regarding the reliability and validity of the retrieved patterns. The review is based on both a-priori (diet scores) and a-posteriori (multivariate statistical analysis) methods. The reports from the existing few studies suggest that the use of both a-priori and a-posteriori pattern analyses in nutrition surveys should be made with consciousness. The suggestion of new statistical techniques for the control of repeatability of dietary patterns is considered essential.

We hypothesized that asthma symptoms and lung function of schoolchildren living in Athens urban area are adversely affected as compared to others living in a rural environment, over a period of 8 years. We recruited 478 and 342 children aged 8-10 years living within a short radius around the urban and rural area monitoring stations, respectively. Respiratory health was assessed by a parent-completed questionnaire in three phases: 1995-1996 (phase-1), 1999-2000 (phase-2), 2003-2004 (phase-3) and by spirometry in phases-1 and 2. Reported asthma and wheeze did not differ in the two areas, whereas cough was more prevalent in the urban area in phase-1. Children from the rural environment had lower levels of percent-predicted forced vital capacity (FVC%) in phase-1 and higher of percent-predicted-forced expiratory flow at mid-FVC (FEF(50)%) in both phases. Independent associations were detected between FVC% as-well-as FEF(50)% and residential area. High FVC% was associated with outdoor systemic athletic activities; there was lower FVC% growth in the urban versus the rural area. Nitrogen dioxide and sulfur dioxide were higher in the urban area, whereas ozone concentrations differed less between the two areas. These results suggest that long-term exposure to urban environment is associated with sub-clinical airway narrowing and slower rate of FVC growth.

The aforementioned findings may suggest that the use of both a priori and a posteriori pattern analysis in nutrition surveys should be made with conscious thought and further research is needed in order to establish robust methodologies to assess the validity of patterns.