Predicting Pediatric Appendicitis using Ensemble Learning Techniques

Abstract

Managing appendicitis in children is still lacking international consensus. This is due to the lack of defined international pediatric appendicitis management recommendations and the unavailability of data-driven studies. In clinical practice, heuristic scoring methods are frequently employed to determine patient urgency. The use of ensemble techniques may improve prediction responses, which is the major goal of this study. This study used an ensemble of classifiers to predict three target variables in 430 children and adolescents aged 0-18 years. The ensemble approaches used in this work are majority voting and weighted averaging. In the Jupyter Python environment, these strategies’ results were confirmed in terms of accuracy, misclassification rate, sensitivity, specificity, precision, f1-score, false negative and positive rates, and Mathew's correlation coefficient. Combining machine learning and deep learning can help in diagnosing and managing pediatric appendicitis, allowing for more tailored medical decisions.