Enhancing Early Stunting Detection: A Novel Approach using Artificial Intelligence with an Integrated SMOTE Algorithm and Ensemble Learning Model

The detection of stunting is a prominent issue in Indonesian healthcare concerning the Sustainable Development Goal of good health and well-being. Stunting poses a significant risk towards children below the age of five. If not treated, stunting may cause symptoms such as a lower cognitive function, lower productivity, a weakened immune system, delayed nerve development, and degenerative diseases. Particularly in regions where stunting is widespread and welfare resources are low, the challenge of detecting children that require treatment is of great importance. Many problems often arise in the diagnostic process, such as the lack of skilled man power to tackle the issue, the lack of experience in medical workers, incompatible anthropometric equipment, and an inefficient medical bureaucracy. For this problem, the implementation of artificial intelligence provides a transformative tool in enhancing medical diagnostic technology. This paper employs and compares the precision, recall, and the f-1 scores of Random Forest, Ada Boost, and Bagging as the performance measure. From the experiment, it is obtained that SMOTE-ENN using Ada Boost classifier and SMOTE using Bagging classifier has the highest F1-Score for minority classes namely stunted and stunting.