A body mass index-based mathematical model on not frequently collected data: the case of Chile
Fecha
2021Autor
Vásquez Pérez, Óscar Carlos
Pérez Casany, Marta
UNIVERSIDAD POLITECNICA DE CATALUNA
Institución
Resumen
The nutritional status is generally measured by body mass index (BMI) being obese, one of the most important non-communicable diseases, a risk factor whose estimation is especially relevant for the development of future health policies. In literature, several authors have addressed its estimation using mathematical/computational models considering frequently collected cross-sectional data to compute the required parameters and the so-called transition probabilities. In this work, we formulate and thereafter implement a non-linear programming (NLP) model to compute BMI transition probabilities, assuming one unit difference changes that remain steady over the time being influenced by the current BMI, sex and age of the person, to estimate a disaggregated characterization of the nutritional status of the over-15 population where, in contrast with other works, the data are not frequently collected. We consider the case of Chile, one of the countries with the highest malnutrition in Latin America and a country where just three surveys are available, through the national health survey, between 2003 and 2017. Several experiments are carried out to find out the best estimation in the goodness-of-fit sense using statistical tests. The obtained results reveal that periods 2003-2010 and 2003-2017 are statistically representative of the actual data, where the fitted transition probabilities are different from one period to the other but they are related. Finally, using these results, the forecasting by 2024 is performed showing a sustained upward trend in terms of BMI and, therefore, in terms of nutritional status, where the worst obese cases would be 12.9% for men aged 55-64 years and 29.3% for women aged 15-24 years.