Early Diagnostics Model for Dengue Disease Using Decision Tree-Based Approaches

Introduction

Dengue is a life threatening disease prevalent in several developed as well as developing nations. This is a virus born disease caused by the breeding of Aedes mosquitoes. Presently, in most regions of the tropics, epidemics are near peak transmission before they are acknowledged and supported as a viral infection.It is mostly too late to implement effective preventive steps that could represent an efficient impact on transmission.To master this situation the surveillance and diagnosis of Dengue should be proactive.The most important concern in developing the dengue diagnostic model is timely prediction of the disease, as its initial symptoms are similar to some other infections (Santosh Kumar et al., 2017; Chopra et al., 2014; Muniaraj, 2014; World Health Organization, 2009,2007). The aim of the survey is to diagnose patients as “Dengue Positive” group or “Dengue Negative’’ on the basis of preliminary symptoms. To address this concern different machine learning techniques for dengue fever classification are used such as Naïve Bayes (NB) classifier (Tu et al., 2009) Decision Tree (DT) (Palaniappan & Awang, 2009), K-Nearest Neighbor (KNN) Technique (Jonsson & Wohlin, 2004), Multilayered Technique, Support Vector Machines (SVM) (Andreeva, 2006) and so along. These techniques can then be assessed based on several performance measures like Accuracy, Precision, Sensitivity, Specificity and Negative rate. This chapter offers a Decision Tree (DT) based dengue diagnostic model. This model can help medical staff to forecast early and accurate diagnosis of dengue using health and medical related data of patients. Moreover, this information can analyze data in different situations and also draws out a pattern of behaviors of patients. The primary aim of Decision Tree (DT) based diagnostic system is to create a model for early detection and diagnosis of dengue disease. Further, three different tree classifiers namely J48, Random tree and REP tree were evaluated and compared with SOM, Logistic regression and Naïve Bayes for precise diagnosis of dengue positive patients. The aims of the proposed chapter by using the real world dengue disease data from the different hospitals located in Delhi (India) are as follows:

• 1.

  Using the real world dengue disease data from the different hospitals located in Delhi (India).

• 2.

  To apply the DT technique for precise detection of dengue positive patients.

• 3.

  To develop a diagnostic model based on machine learning techniques for early detection and diagnosis of dengue disease and furthermore for the assistance of physicians.

• 4.

  To validate the results of the proposed diagnostic model.