Machine Learning Method to Predict the Toddlers’ Nutritional Status

Abstract

Malnutrition is one of the leading health problems experienced by toddlers in various countries. Based on the 2022 Indonesian Nutritional Status Survey results, malnutrition in children under five in Indonesia is higher than the average malnutrition in Africa and globally. Therefore, a way is needed to predict the nutritional status of children under five early and quickly so that the Government (through District Health Office) can immediately provide the necessary treatment. This study aims to predict or classify the toddlers’ nutritional status based on age, body mass index (BMI), weight, and body length using various machine learning (ML) methods, namely naïve Bayes, linear discriminant analysis, decision tree, k-nearest neighbor, random forest, and support vector machine. The predictive performance of each ML method was evaluated based on accuracy, sensitivity, specificity, the area under curve, and Cohen's Kappa coefficient. The test results show that the RF method is the most recommended for predicting toddlers' nutritional status. The study's contribution is to obtain information about toddlers' nutritional status easier.