Classification of weight status using anthropometric and clinical indicators

Abstract

Obesity is a chronic disease that became a crisis on adult and child health worldwide, especially in Mexico where about 70 % of Mexican adults are overweight. Although several efforts have been made in order to take control over the problem, obesity as a disease is complicated to monitor and the solutions have had a limited success. In the other hand, disorder eating behaviors have been reported as an increasing health problem. Both clinical and anthropometric measurements can help on identify weight condition and in this paper we present an analysis of some of these attributes on a database devoted to identify the weight status of the subject. The data was collected from 600 records of a clinical laboratory in Mexico City and the attributes included cholesterol, LDL, HDL, body mass index (BMI) and waist measurement. We applied two classifiers to find out the way the classes are distributed and the possible relationship between the two categories of attributes. The classes were grouped in three categories (thinness, normal and obesity) and three levels of risk according to the experts for a total number of 28 classes. The results showed issues on severe thinness, medium and morbid obesity classes as well a strong impact of BMI on the classification of the obesity classes. The classification can be seen as an aid tool to help managing weight problems by an integral subjectcentered strategy.