Machine learning-based stunting diagnosis: utilisation of real-time anthropometric data from IoT devices with the XGBoost algorithm

Abstract

Stunting, or chronic malnutrition, is a global public health issue that requires early and accurate intervention. Traditional diagnostic methods are often prone to manual measurement errors and are time-consuming. This study proposes an innovative approach to early diagnosis of stunting by integrating Internet of Things (IoT) technology for real-time anthropometric data collection and Extreme Gradient Boosting (XGBoost) algorithms for predictive analysis. The developed system utilises IoT-based anthropometry devices to automatically measure children's height, minimising human error and providing real-time data. This raw data is then processed into Z-scores, which become the main features (input) for the Machine Learning model. The XGBoost model was chosen for its ability to handle imbalanced data and provide high classification accuracy. In testing, the XGBoost model demonstrated superior performance in classifying nutritional status (Normal, Stunting) with key metrics such as Accuracy (>95%) and a high F1-Score, surpassing other conventional machine learning models. By combining precise measurements from IoT and the predictive power of XGBoost, this application offers a fast, accurate, and measurable diagnostic solution. This innovation has great potential to support government programmes and health workers in continuously monitoring child growth and making timely nutritional intervention decisions.