Application of Fuzzy Cognitive Map for geospatial dengue outbreak risk prediction of tropical regions of Southern India
Article type: Research Article
Authors: Jayashree, L.S.a | Lakshmi Devi, R.b | Papandrianos, Nikolaosc | Papageorgiou, Elpiniki I.d; *
Affiliations: [a] Computer Science Engineering, PSG College of Technology, Coimbatore, India | [b] Affiliations as Computer Applications, GSS Jain College, Chennai, India | [c] Nursing Department, Technological Educational Institute of Central Greece, Lamia 35100, Greece | [d] Computer Engineering Department, Technological Educational Institute of Central Greece, Lamia 35100, Greece
Correspondence: [*] Corresponding author: Elpiniki I. Papageorgiou, Computer Engineering Department, Technological Educational Institute of Central Greece, Lamia 35100, Greece. E-mail: [email protected].
Abstract: BACKGROUND: Dengue is one of the serious mosquito-borne diseases predominantly caused by the bites of infected Aedes mosquitoes. The global incidence of dengue has significantly increased in the last few years. According to the World Health Organization (WHO), there exist 390 million dengue fever cases worldwide. In India, the number of cases of dengue was doubled during 2014 and 2015. Different geographical regions possess different levels of dengue outbreak risk depending on the meteorological, socioeconomic and lifestyle parameters such as cleanliness around the house, water storage practice, use of mosquito repellent, family history etc. Hence, a geospatial categorization helps in exercising timely prevention mechanism by the public health service personnel before the real outbreak occurs. METHODS: The presented work attempts to classify the dengue outbreak risk of the given geographical region using Fuzzy Cognitive Map (FCM). In this research study, medical experts contributed to construct the FCM model for assessing dengue risk. This model was trained using data-driven nonlinear Hebbian learning (DD-NHL) algorithm for 71 data samples collected from suburban areas of Chennai city, one of the tropical regions of Southern India. The new contribution of this research work is the application of the FCM methodology with its Hebb-based learning capabilities in the specific application case study. The proposed FCM model classifies the dengue outbreak risk of the given region into three categories, namely high, moderate and low. RESULTS: A large number of experiments were conducted with diverse configurations and different learning parameters of Hebb-based learning algorithms, as well as with different architectures of artificial neural networks (mainly for comparison purposes). Through the comparative analysis, also the standard machine learning based classifiers such as Multilayer Perceptron, Support Vector Machine (SVM), Decision Tree, Naive Bayesian classifier, etc. were used. CONCLUSIONS: The accuracy of the proposed FCM-based classification approach is much better than the benchmark machine learning algorithms, which show deficiency working with small datasets and without be able to use experts’ knowledge. Dengue is one of the serious mosquito-borne diseases predominantly caused by the bites of infected Aedes mosquitoes. It causes a wide spectrum of illness ranging from mild asymptomatic illness to severe fatal Dengue Hemorrhagic Fever/Dengue Shock Syndrome (DHF/DSS). The global incidence of dengue has significantly increased in the last few years. According to the World Health Organization (WHO), there exist 390 million dengue fever cases worldwide. In India, number of cases of dengue was doubled during 2014 and 2015. Different geographical regions possess different levels of dengue outbreak risk depending on the meteorological, social and lifestyle factors such as cleanliness around the house, water storage practice, use of mosquito repellent, family history etc. The presented work attempts to classify the dengue outbreak risk of the given geographical region
DOI: 10.3233/IDT-180330
Journal: Intelligent Decision Technologies, vol. 12, no. 2, pp. 231-250, 2018
