Contents lists available at ScienceDirect New Horizons in Translational Medicine journal homepage: www.elsevier.com/locate/nhtm Research article PSO-ANN based diagnostic model for the early detection of dengue disease Shalini Gambhir a , Sanjay Kumar Malik a , Yugal Kumar b, ⁎ a Department of Computer Science and Engineering, SRM University, Haryana, India b Department of Computer Science and Engineering, JUIT, Himachal Pardesh, India ARTICLE INFO Keywords: Classification Neural network Decision tree Naive Bayes CART PSO ABSTRACT Large numbers of machine learning approaches have been developed for analysis of medical data in recent years. These approaches have also proved their significance through accurate and earlier diagnosis of diseases. The objective of this work is to develop a diagnostic model for earlier diagnosis of dengue disease. Dengue fever is spread through the bite of the female mosquito (Aedes aegypti). The symptoms of this fever are similar to other fever such as that of Viral influenza, Chikungunya, Zika fever, and so on. However, in this fever, human life can be at risk due to severe depletion of blood platelets. Therefore, early diagnosis of dengue disease can help in protecting human lives by making a preventive move before it turns into an infectious disease. In this work, an effort is made to develop a PSO-ANN based diagnostic model for earlier diagnosis of dengue fever. In the pro- posed model, PSO technique is applied to optimize the weight and bias parameters of ANN method. Further, PSO optimized ANN approach is used to detect dengue patients. The effectiveness of the proposed model is evaluated based on accuracy, sensitivity, specificity, error rate and AUC parameters. The results of the proposed model have been compared with other existing approaches like ANN, DT, NB, and PSO. It is observed that the proposed diagnostic model is a proficient and powerful model for more accurate and earlier detection of dengue fever. 1. Introduction Dengue is a mosquito-based viral disease that can quickly spread in favorable climatic condition. This communicates through female mos- quito named ‘Aedes aegypti’. The main reason behind the widespread prevalence of dengue disease over the tropics is due to variations in rainfall, temperature, and unplanned rapid urbanization. In recent years, dengue cases have grown up rapidly around the world, however, the actual numbers of dengue cases are either never reported or sometimes classified inaccurately. According to WHO report, every year, 390 million dengue infections are reported in the entire world, out of this 96 million are clinically reported with the severity of disease [1]. The other study, on the occurrence of dengue disease, indicates that dengue viruses can infect 3.9 billion people in 128 countries [2]. The number of cases registered for dengue is increased from 2.2 million (in 2010) to 3.2 million (in 2015). Dengue is one of most fatal and wide- spread viral infection in the world today. It is an increasingly prevalent tropical virus infection with significant morbidity and fatality rate [3]. Dengue infection has been recognized to be endemic in India for over two centuries as a benign and self-limited disease. In recent years, the disease has shifted its course manifesting in the severe form of DHF and with increasing frequency of outbreaks [4]. Dengue infection in a previously non-immune host produces a principal response of anti- bodies characterized by a slow and low-titer antibody response. IgM antibody is the first immunoglobulin Isotype to appear. In a suspected case of dengue, the presence of anti-dengue IgM antibody suggests re- cent infection. Anti-dengue IgM detection using enzyme-linked im- munosorbent assay (ELISA) represents one of the most important ad- vances and has become an invaluable instrument for routine dengue diagnosis [5]. In recent years, various decision support systems and diagnostic models have been developed for improving experiences and abilities of physicians to accurate detection and diagnosis of diseases. It is observed from the recent research trends that artificial neural networks have been widely used in the field of medical data mining and number of decision support systems have been developed with the help of ANN due to its ability of prediction, parallel operation and Adaptivity [6–13]. The multilayer neural networks (MLNNs) have been success- fully used in replacing conventional pattern recognition methods for the disease diagnosis systems and it can be back-recognized as a http://dx.doi.org/10.1016/j.nhtm.2017.10.001 Received 19 July 2017; Received in revised form 12 September 2017; Accepted 8 October 2017 ⁎ Corresponding author at: Department of Computer Science and Engineering, JUIT, Himachal Pardesh, India. E-mail addresses: shalinigambhir.21@gmail.com (S. Gambhir), skmalik9876@gmail.com (S.K. Malik), yugalkumar.14@gmail.com (Y. Kumar). Abbreviations: ANN, Artificial Neural Network; ANFIS, Adaptive Neuro-Fuzzy Inference System; AUC, Area Under the Curve; BPSO, Binary Particle Swarm Optimization; CART, Classification and Regression Tree; DHF, Dengue Hemorrhagic Fever; DT, Decision Tree; GA, Genetic Algorithm; MLNN, Multilayer Neural Network; MLP, Multilayer Perceptron; NB, Naive Bayes; PSO, Particle Swarm Optimization; SVM, Support Vector Machine; WHO, World Health Organization New Horizons in Translational Medicine xxx (xxxx) xxx–xxx 2307-5023/ © 2017 European Society for Translational Medicine. Published by Elsevier Ltd. All rights reserved. Please cite this article as: Gambhir, S., New Horizons in Translational Medicine (2017), http://dx.doi.org/10.1016/j.nhtm.2017.10.001