Modelo Predictivo para la Desnutrición Infantil.

Abstract

Abstract: Child malnutrition in Ecuador is a priority public health problem due to its multidimensional effects on children's physical, cognitive, emotional, and social development, causing disadvantages that extend throughout the life cycle. Therefore, early identification through predictive tools represents a strategic aid for health decision-making. This study aimed to develop a predictive model for child malnutrition using demographic and health records of children from 2022 and 2024, provided by the National Institute of Statistics and Censuses (INEC). After analyzing, visualizing, cleaning, completing missing values, and normalizing the variables, the model was trained using three machine learning algorithms: Logistic Regression, Random Forest, and XGBoost. According to the results obtained, the XGBoost algorithm clearly outperformed the other two in terms of performance and accuracy. Among all the variables in the data set, both from a medical point of view and from the correlation result, age, weight and height in children are key for the model to be able to predict childhood malnutrition and to be able to plan, intervene and monitor the health of the infant in advance.