DEVELOPMENT OF CLINICAL DECISION SUPPORT ENSEMBLE-BASED SYSTEM FOR THE PREDICTION OF CHILDHOOD KILLER DISEASES IN RESOURCE-LIMITED SETTINGS

ABSTRACT

Childhood killer diseases like malaria, pneumonia, diarrhoea are still the major causes of death among children under the age of five in resource-limited settings especially in sub-Saharan Africa. Poor healthcare infrastructure, shortage of skilled manpower and absence of smart diagnostic tools at Primary Health Care (PHC) level are often a barrier to early diagnosis. This paper gives the design of an ensemble-based Clinical Decision Support System (CDSS) in prediction of childhood killer diseases. The suggested system combines several machine learning algorithms, such as Support Vector Machine (SVM), Random Forest (RF), Naïve Bayes (NB), K-Nearest Neighbors (KNN) and Decision Tree (DT) with a majority voting ensemble methodology to enhance prediction accuracy and robustness. The model is trained on retrospective clinical data (3-5 years) and tested on prospective data (12 months) where seasonal differences in disease patterns are accounted. The system is in line with WHO IMCI and iCCM to provide clinical relevancy. Accuracy, precision, recall, and F1-score-based performance evaluation show that the ensemble model is superior to the single classifiers. The suggested CDSS will improve the accuracy of diagnosis, assist frontline health workers, and decrease the under-five death rates in limited resource settings.

Keywords: Clinical Decision Support System, Ensemble Learning, Machine Learning, Childhood Diseases, Predictive Modelling, Primary Health Care.