Naive Bayes Analysis for Nutritional Fulfillment Prediction in Children

Abstract

Stunting in children remains a significant global health challenge, particularly in low- and middle-income countries. Addressing this issue requires an effective approach to predicting and preventing inadequate nutritional fulfillment. This study uses the Naïve Bayes approach to forecast nutritional needs for children's growth and development, providing practical information for stunting prevention efforts. The data used were sourced from 174 infant and toddler examinations at the Puskesmas Lawang, involving eight key attributes: gender, age, weight, height, head circumference, pre-screening, vision tests, and nutritional status. Key performance metrics were evaluated to validate the model's predictive capabilities, including accuracy, precision, recall, and F1-score. Six test scenarios were conducted using different percentages of training data (90%, 80%, 70%, 60%, 50%, and 40%) to evaluate the reliability of the Naïve Bayes method. Results indicated that the highest accuracy of 78.84% was achieved in the sixth test scenario. The third test scenario produced the highest precision at 97.5%, while the highest recall (100%) was observed in the first three scenarios. The highest F-measure of 90.3% occurred in the fourth scenario. These results suggest the algorithm's potential for early detection to decrease the number of stunting children. The study’s implications are twofold: practically, the model can be integrated into health monitoring systems to assist healthcare professionals and policymakers in designing more effective nutrition programs; theoretically, it highlights the adaptability of Naive Bayes for handling complex, multi-dimensional health data.

Keywords

• artificial intelligence
• machine learning
• early detection of child physical growth
• nutritional fulfillment
• health issues