Evaluation of Dengue Model Performances Developed Using Artificial Neural Network and Random Forest Classifiers

Abstract

Abstract Dengue is one of the endemic diseases in Indonesia. Dengue is being suffered by many people, regardless of their gender and age. Therefore, research about dengue based on dengue patients’ data was conducted. There was a lot of information written in that data regarding the corresponding patients and the dengue they had suffered, such as gender, age, how long the patients were hospitalized, the symptoms they experienced, and laboratory characteristics results. Diagnosis of each of the corresponding patients based on their symptoms and laboratory characteristics results were also written in that data. The diagnoses were classified into three different clinical degrees according to the severity level, which is DF as the mild level, DHF grade 1 as the intermediate level, and DHF grade 2 as the severe level. In this research, data of the patients on the third day of being hospitalized was analyzed, because, on the third day, dengue is entering a critical phase. The objectives of this research were: to evaluate the performance of the models that were used to predict the correct class within the given dataset developed using Artificial Neural Network (ANN) classifier and Random Forest (RF) classifier separately, and to find a classifier that yielded the best performance. The results obtained from this research will be used in the development of a Machine Learning model that can predict the clinical degree of dengue in the critical phase, if the laboratory characteristics results are known, using a classifier that yielded the best performance.