Machine Learning Classifier Algorithms for Predicting Malnutrition Among Under Five Children in Asia

Abstract

Malnourished children may have serious health issues. Furthermore, doctors often struggle to pinpoint the underlying causes of their patients' ailments, leading them to perform surgeries that may not be appropriate for all children. This is a frequent reason why children die. As a result, undernourished children are put in grave danger. Therefore, the primary goal of our research is to use AI to forecast the starvation status of children aged 0 to 5 in Asia. We looked for active research papers from 2010 to 2020 that accepted our point of view, consolidated the data, and attempted to identify benefits and downsides. Like I said before, we used an acceptable open-source dataset for this. They also studied several articles to gain an understanding of the benefits and drawbacks of ML techniques. Eight common ML classifiers Random Forest, Support Vector Machine, K-Nearest Neighbors, Logistic Regression, Bernolli Naive Bayes, Complement Naive Bayes, Decision Tree, and Gradient Boosting predict malnutrition in children under 5 with excellent accuracy. Finally, they searched for algorithms with the highest accuracy scores. Logistic Regression and K-Nearest Neighbors performed best, with train accuracy of 1.000 and 0.98 and success rates of 95.34% and 93.02%, respectively. Furthermore, the application of logistic regression classification indicated a very significant capacity to detect differences. They looked at eight machine learning algorithms to discover which one was the most successful. Among them, Logistic Regression and K-Nearest Neighbors do very well. My aim is to alleviate the future suffering of malnourished children. My next research will focus on Bangladesh's highland and coastal areas, which have poor educational levels and a high risk of child marriage.