Dengue Outcome Prediction in the Dhaka Region: A Classification Approach Using Demographic and Diagnostic Features

Abstract

The purpose of this study is to predict the course of dengue fever in the Bangladeshi city of Dhaka by using a classification approach that makes use of diagnostic and demographic information. The first machine learning models, Support Vector Machine (SVM) and kNearest Neighbors (KNN), were applied using data samples that included demographic and medical variables. The results showed predicted accuracy levels of 61% and 66%, respectively (referenced dataset). Later advances included standardizing data scaling and tuning hyperparameters. These yielded significant gains in performance; SVM reached an astounding 93%, while KNN exceeded expectations at almost 100%. The study also presented a Meta-hybrid model that uses a stacking-classifier in a machinelearning ensemble. This novel method used the advantages of separate models to produce a remarkable 92% prediction accuracy for the dengue outcome of individuals in particular conditions. The results emphasize