Machine learning algorithms in the detection of anemia in children under 5 years of age in southern Peru

Arroyo-Paz A.; Aleman-Gonzales L.; Ingaluque-Arapa M.; Tapia-Catacora P.; Rodriguez-Aburto C.; Larico-Uchamaco G.R. · Congreso Internacional de Innovacion y Tendencias en Ingenieria, CONIITI (2024) · DOI: 10.1109/coniiti64189.2024.10854794

Resumen

The present study focuses on the challenge of identifying anemia in children under 5 years of age. in the departments of Cusco, Moquegua, Puno, Tacna and Arequipa, considered as the southern zone of Peru. The importance of early detection of this disease is highlighted due to the irreversible consequences. The main objective is to apply and compare the various Machine Learning algorithms for the early detection of this disease. The aim is to evaluate the effectiveness of the KNN, Decision Tree and Naïve Bayes models, in order to identify the most accurate algorithm. A dataset from the Information System of the Nutritional Status of Children and Pregnant Women Peru (SIEN) was used, including variables such as sex, age in months, weight, height, hemoglobin, among others. Data cleaning and processing techniques were applied to prepare them for the use of the different Machine Learning algorithms. The results obtained show high accuracy rates of the algorithms used, the algorithms with the highest precision were Decision Tree and KNN with 99.21% and 96.56% respectively. In addition, Naïve Bayes showed remarkable performance with an accuracy of 94.94%. Based on the comparisons of accuracy rates, it is concluded that the most effective algorithm for the early detection of anemia in children in southern Peru is Decision Tree.

Instituciones: Universidad Nacional del Altiplano; Universidad Nacional de Cañete; Universidad Nacional del Callao