The efficiency of artificial neural network (ANN) for diagnosis of obesity and hypertension

Obesity and hypertension are health problems in any society. The aim of this study was to evaluate thesensitivity, specificity and accuracy of artificial neural network (ANN) for the diagnosis of obesity andhypertension. For this study, demographic information about ۵۵۰ students aged ۷-۱۸ years was recorded inthe ANN program. The recorded demographic information consisted of ۱۱ input variables and ۳ outputvariables. Input variables included age, sex, weight, height, waist circumference, body mass index, waist-toheightratio, abdominal obesity, physical activity, genetics, and unhealthy eating behaviors, while outputvariables included obesity, systolic blood pressure, and diastolic blood pressure. In this study, Levenberg-Marquardt and Conjugate Gradient algorithms were used to training the network. The results showed that theselected neural network with Levenberg-Marquardt algorithm had ۱۷ hidden neurons in the diagnosis ofobesity and high diastolic blood pressure, while in the diagnosis of high systolic blood pressure it had ۱۵hidden neurons. Based on the results of the study, the sensitivity, specificity and accuracy of the network inthe diagnosis of diastolic blood pressure were ۰.۸۱۲۳, ۰.۹۹۱۵ and ۰.۹۷۱۳, respectively. While these valueswere ۰.۹۶۷۲, ۰.۹۹۶۲ and ۰.۹۸۱۸ for obesity and ۰.۸۵۵۹, ۰.۹۹۱۲ and ۰.۹۸۴۳ for systolic blood pressure,respectively. Based on the results of the present study, it can be concluded that ANN designed to diagnoseobesity, systolic and diastolic blood pressure with equal accuracy of ۹۶%, ۸۵% and ۸۱%, respectively.Therefore, it can be said that ANN program has high efficiency in diagnosing obesity and hypertension.

کلیدواژه ها:

Artificial Neural Network ، Health ، Obesity ، Hypertension ، Efficiency ، Iran