Prediction of Dengue using Recurrent Neural Network

India and various developing nations where the population is high, healthcare is one of the major challenges to deal with. The medical resources that are made available by the government cannot cope up with the high population. The rural areas are affected the most due to lack of a proper medical infrastructure by public health sector. A large chunk of population cannot afford private hospitals as they belong to lower middle class or below poverty line group. Therefore, the load on public health sector has increased manifold. In case of an epidemic, the death rate is high as most of the population have no access to medical resources and thus timely treatment. As a large chunk of population of our country lives in rural areas, it becomes a dire need to arrange for a proper medical infrastructure in case an epidemic occurs. Dengue being one of the most occurring epidemics in our country as well as the reason for a high fatality rate, needs to be predicted well in future so that medical resources can be arranged on time and thus reduce the fatality rate. Although dengue does not require a specific line of treatment or prolonged, it is essential to be detected and treated on time. Our system aims at predicting epidemic (dengue), so that the health sector can arrange for necessary resources before time. The system will make use of Recurrent Neural Network for prediction. Data used for prediction will include the climatic conditions, pollution and the statistics of patients diagnosed with dengue in the previous years. The data has been taken from various government websites as well as healthcare centers. The model will learn using this data and predict the possibility of an outbreak if similar climatic conditions occur in future. The result of the possibility of an outbreak will be shown by highlighting the regions where dengue might occur using google heatmaps.