Associating Obesity to Chronic Conditions through Machine Learning Techniques: A Mexican Case

Abstract

Eating disorders do not have a single cause, so the prediction of suffering from them is a complex issue. The economic cost associated with the treatment of diseases and disabilities derived from overweight and obesity is expensive in relation to the cost of prevention. This research aims to evaluate analysis techniques to identify the relationship between obesity in Mexico and several factors (variables) such as habits and physical conditions, using information technologies. Artificial Intelligence is a resource that can help in complex multifactorial issues, such as the identification of risks in different fields or combination of them. Methodologically, the use of Mexican public databases and expert knowledge is proposed to feed the Artificial Intelligence system to achieve early prediction and/or diagnosis of obesity condition. Throughout the bibliographic reference, the benefit of the use of these technologies for predictive analysis in the health arena is identified, being an opportunity to make this technology available to many people showing the impacts and consequently, favoring prevention. It is demonstrated how the Random Forest model presents a better performance with respect to the other techniques for the classification problem presented.