Mochammad Rizky Kusuma, Mukh Taofik Chulkamdi, Sri Lestanti

Implementasi Metode Adaptive Neuro-Fuzzy Inference System (ANFIS) untuk Prediksi Status Gizi Balita Studi Kasus Wilayah Kabupaten Blitar

Introduction

Implementasi metode adaptive neuro-fuzzy inference system (anfis) untuk prediksi status gizi balita studi kasus wilayah kabupaten blitar. Model prediksi status gizi balita (stunting & wasting) di Blitar menggunakan ANFIS. Akurat mendeteksi dini, membantu tenaga kesehatan intervensi cepat, percepat penurunan stunting.

Abstract

This study aims to develop a predictive model for the nutritional status of toddlers using the Adaptive Neuro-Fuzzy Inference System (ANFIS) method based on anthropometric data. According to WHO data (2023), global nutritional problems, such as wasting and stunting, are alarming, with 45 million toddlers experiencing wasting and 149 million experiencing stunting. In Indonesia, the prevalence of stunting was recorded at 24.4%, higher than the WHO threshold of 20%. In Blitar Regency, the prevalence of stunting also increased from 14.3% (2022) to 20.3% (2023), a contributing factor being the manual recordingsystem at community health centers (Puskesmas) and integrated health posts (Posyandu). This study used data from 5,000 toddlers from the Kanigoro Community Health Center and Gogodeso Integrated Health Post (Posyandu), with 70% of the data allocated for training and 30% for testing. Model evaluation was conducted using three metrics: Mean Squared Error (MSE), Root Mean Square Error (RMSE), and Mean Absolute Error (MAE). The evaluation results demonstrated the best prediction accuracy in the MSE for weight/age, height/age, and weight/height, indicating stable data vriation and sensitivity to outlier detection. This prediction system was implemented using MATLAB GUIDE, making it practical for use by healthcare professionals. The results of this study can support efforts to accelerate stunting reduction through faster and more accurate predictions of toddler nutritional status

Review

The paper "Implementasi Metode Adaptive Neuro-Fuzzy Inference System (ANFIS) untuk Prediksi Status Gizi Balita Studi Kasus Wilayah Kabupaten Blitar" addresses a highly relevant and critical public health issue: the prediction of toddler nutritional status, particularly in the context of stunting and wasting. The authors' choice to employ the Adaptive Neuro-Fuzzy Inference System (ANFIS) is commendable, as it offers a robust approach for handling complex, non-linear relationships inherent in health data. The large dataset of 5,000 toddlers from a specific region in Blitar Regency provides a solid foundation for training and testing, and the reported implementation in MATLAB GUIDE suggests a practical and accessible tool for healthcare professionals, aligning well with the goal of supporting stunting reduction efforts. The methodology outlines a clear approach using anthropometric data, with a standard 70/30 split for training and testing, and evaluation metrics including Mean Squared Error (MSE), Root Mean Square Error (RMSE), and Mean Absolute Error (MAE). The abstract states that the evaluation demonstrated "best prediction accuracy in the MSE for weight/age, height/age, and weight/height, indicating stable data variation and sensitivity to outlier detection." While this claim highlights the model's performance in these crucial indicators, the abstract would benefit from providing the actual numerical values of these evaluation metrics. Quantifying the MSE, RMSE, and MAE would significantly strengthen the evidence for the model's predictive power and allow for a more objective assessment of its practical utility. While the study presents a promising application of ANFIS, further details would enhance its impact. Future work should consider a comparative analysis with other established machine learning or deep learning models to rigorously benchmark ANFIS's performance and potentially identify even more optimal solutions. Additionally, exploring the generalizability of this model to other regions in Indonesia or globally, and discussing the ethical considerations and user acceptance of such a predictive system in real-world Puskesmas and Posyandu settings, would be valuable. Despite these minor points, this research makes a significant contribution by offering a data-driven, practical solution to a pressing public health challenge, with clear potential for real-world application in accelerating stunting reduction initiatives.

Full Text

Comments

You need to be logged in to post a comment.