Determining Skinfold Thickness through Artificial Neural Networks

Abstract

Skinfold thickness values, which are used in many studies to evaluate the nutritional status of disabled people, are measured by a caliper. The usage of this measurement device in disabled people, and especially children with autism, is very difficult. The aim of this study is to find a solution to this problem by using a model that produces high accuracy predictions without needing a caliper. In this study, an artificial neural network application was realized for the determination of skinfold thickness in individuals. In order to train the artificial neural network (ANN) model, the body weights, body mass indexes, waist circumferences, abdomen circumferences, and hip circumferences of 400 children and adolescents between the ages of 6 and 18 were measured. 138 sets of the data collected from the 400 children and adolescents were selected to be used with the ANN, and a data set was formed. 70% of this data set (96 in number) was used for the training of the model, 15% (21 in number) was used for testing the model, and the remaining 15% was used for approving the model. As a result of the study, a prediction with 97.7% accuracy was obtained, and a highly close relationship between the output of the artificial neural network and the target was found.