Food consumption habits, gestational age and birth weight are predictive for children with excess weight: An analysis based on artificial neural network

Abstract

The relationship between body weight gain and the onset of obesity is linked to environmental and behavioral factors, and may be dependent on biological predisposing. Artificial neural networks are useful predictive tools in the field of artificial intelligence, and can be used to identify risk factors related to obesity. The aim of this study is to establish, based on artificial neural networks, a predictive model for overweight/obesity in children based on the recognition and selection of patterns associated with birth weight, gestational age, height deficit, food consumption, and the physical activity level, TV time and family context. Sample consisted of 149 children (72 = eutrophic and 77 = overweight/obese). Collected data consisted of anthropometry and demographic characteristics, gestational age, birth weight, food consumption, physical activity level, TV time and family context. The gestational age, daily caloric intake and birth weight were the main determinants of the later appearance of overweight and obesity. In addition, the family context linked to socioeconomic factors, such as the number of residents in the household, had a great impact on excess weight. The physical activity level was the least important variable. Modifiable risk factors, such as the inadequate food consumption, and non-modifiable factors such as gestational age were the main determinants for overweight/obesity in children. Our data indicate that, combating excess weight should also be carried out from a social and preventive perspective during critical periods of development, such as pregnancy, lactation and early childhood, to reach a more effective strategy to combat obesity and its complications in childhood and adult life.