Morbidity Rate Prediction of Dengue Hemorrhagic Fever (DHF) Using the Support Vector Machine and the Aedes aegypti Infection Rate in Similar Climates and Geographical Areas

Abstract

BackgroundIn the past few decades, several researchers have proposed highly accurate prediction models that have typically relied on climate parameters. However, climate factors can be unreliable and can lower the effectiveness of prediction when they are applied in locations where climate factors do not differ significantly. The purpose of this study was to improve a dengue surveillance system in areas with similar climate by exploiting the infection rate in the Aedes aegypti mosquito and using the support vector machine (SVM) technique for forecasting the dengue morbidity rate.Methods and FindingsAreas with high incidence of dengue outbreaks in central Thailand were studied. The proposed framework consisted of the following three major parts: 1) data integration, 2) model construction, and 3) model evaluation. We discovered that the Ae. aegypti female and larvae mosquito infection rates were significantly positively associated with the morbidity rate. Thus, the increasing infection rate of female mosquitoes and larvae led to a higher number of dengue cases, and the prediction performance increased when those predictors were integrated into a predictive model. In this research, we applied the SVM with the radial basis function (RBF) kernel to forecast the high morbidity rate and take precautions to prevent the development of pervasive dengue epidemics. The experimental results showed that the introduced parameters significantly increased the prediction accuracy to 88.37% when used on the test set data, and these parameters led to the highest performance compared to state-of-the-art forecasting models.ConclusionsThe infection rates of the Ae. aegypti female mosquitoes and larvae improved the morbidity rate forecasting efficiency better than the climate parameters used in classical frameworks. We demonstrated that the SVM-R-based model has high generalization performance and obtained the highest prediction performance compared to classical models as measured by the accuracy, sensitivity, specificity, and mean absolute error (MAE).

An Innovative Big Data Predictive Analytics Framework over Hybrid Big Data Sources with an Application for Disease Analytics

“…In contrast, Kesorn et al [13] observed that only weather variables would not be sufficient for precisely forecasting dengue outbreaks. They introduced "mosquito infection rate" as a new feature, and used six different machinelearning models for comparison in two different scenarios (namely, with vs. without the new feature).…”

Use of artificial intelligence in infectious diseases

“…Applying SVM with the radial basis function (RBF) kernel, scientists were able to forecast the high morbidity rate and take precautions to prevent such cases to happen. The parameter that was able to reach a high level of accuracy was not linked to climate but to the infection rate of the mosquitoes that transmit dengue virus (Kesorn et al, 2015). It most cases of infectious diseases the success transmission blockade is usually linked to outreach (Saybani et al, 2016).…”

A comparison of three data mining time series models in prediction of monthly brucellosis surveillance data

“…There are several time series model methods that can be used to forecast the incidence of infectious diseases and have been widely utilized in this context (Aloufi et al, 2016;Kane et al, 2014;Kesorn et al, 2015;Wu et al, 2017;Zhang et al, 2016Zhang et al, , 2014. Recently, machine learning-based time series models have been effectively used for modelling incidence of infectious disease and forecasting problems in public health studies including support vector machines (SVMs) (Kesorn et al, 2015;Kisi, Parmar, Soni, & Demir, 2017;Zhang et al, 2014), random forest (RF) (Kane et al, 2014;Wu et al, 2017) and multivariate adaptive regression splines (MARSs) (Kisi et al, 2017). These techniques account for the non-linear effects of predictors (which are usually considered as linear in traditional models or where possible there is limitation in the number of non-linear effects in traditional models due to issues such as identifiability) as well as all interactions between predictors which make machine learning methods powerful tools for forecasting.…”