Body Mass Index Estimation by Using an Adaptive Neuro Fuzzy Inference System

In the last few decades, the early diagnosis and prediction of disease, particularly the diseases related to Body Mass Index (BMI), have increased dramatically. In this regard, the soft computing techniques, such as neural networks, machine learning, Fuzzy Inference Systems (FIS) etc. are the principle elements in these studies. The Artificial Neural Network particular form with the hybrid intelligent system called Adaptive Neuro-Fuzzy Inference System (ANFIS) was chosen and applied in this study. ANFIS benefits from the ANN’s superior learning algorithms and FIS’ excellent estimation functions. The main aim of the current study is to examine the performance of ANFIS in order to estimate the BMI by using correlated explanatory variables with the response variable BMI. Two ANFIS models with the same attributes and structures were developed separately for both gender data sets. The low RMSE results of both models (female RMSE=1.914 and male RMSE=1.817) revealed that these models are biologically acceptable and provide methods for predicting the BMI based on five factors, which have acceptably strong correlations with BMI values. Consequently, it is suggested that obesity could be prevented if the changeable risk factors with various lifestyle modifications could be controlled.