Automatic Prediction System of Dengue Haemorrhagic-Fever Outbreak Risk by Using Entropy and Artificial Neural Network

Predicting Dengue Haemorrhagic Fever outbreak is obviously urgent in order to control and prevent a widespread of the fever in advance. However, the prediction of Dengue Haemorrhagic Fever outbreak needs the analysis from experts which is inconvenient and costly. An automatic prediction system should be developed. This paper proposes an automatic prediction system of Dengue Haemorrhagic-Fever outbreak risk by using entropy technique and artificial neural network. In this system, the information extraction is preprocessed prior to the prediction in order to reduce data redundancy and retain only those relevant data. First, the external factors such as temperature, relative humidity, and rainfall are considered during the information extraction. Then, a supervised neural network is deployed to predict the possible risk of Dengue Haemorrhagic Fever outbreak. To evaluate the performance of proposed system, the experiments based on the condition of weather data and Dengue Haemorrhagic Fever cases from January 1999 until December 2007 were conducted. Our prediction achieves 85.92% accuracy compared to the actual data.