Improved Prediction of Dengue Outbreak Using the Delay Permutation Entropy

Abstract

Climate is an important contributing factor in the outbreak and spread of dengue fever because it strongly affects the density and distribution of the mosquitoes that carry dengue. However, dengue forecasting models based solely on local weather factors have had limited success. Hence, this paper proposes a novel dengue outbreak detection method that relies on delay permutation entropy (DPE) features extracted from daily weather data. Using data from Hong Kong between 2004 and 2015, 4383 daily DPE features have been extracted and analysed using machine learning techniques. The analysis results show that there is a strong correlation between dengue cases and rainfall DPE features. A comparison with predicted results generated from average monthly weather data, shows that the dengue outbreak detection results based on the DPE features, are up to 11% more accurate than those predicted from the monthly average data.