Predicting the Recovery Rate of COVID-19 Using a Novel Hybrid Method

COVID-19 pandemic and its transformation into a global health emergency have further highlighted the need to design an intelligent system that can analyze the growing information of such an epidemic. At the beginning of the outbreak, due to the lack of information about the factors affecting the recovery of people compared to the information of infected or dead ones, designing a system based on such biased data that can accurately predict the recovered cases is very important. Such a system can help health officials make decisions in complex situations and reduce public anxiety. In this study, a new hybrid structure for predicting the recovery rate of COVID-19 is presented. This structure, which is also suitable for binary classification tasks in medical applications, is a combination of algorithms with high sensitivity and specificity criteria. The COVID-19 dataset provided by Johns Hopkins University was used to evaluate the performance of the model. We used the data from the first 43 days of outbreak in 160 different regions around the world. The accuracy, sensitivity and specificity of the model on test set were 84.54%, 78.53% and 88.73%, respectively. These promising results show that the model can be used to analyze other medical data on which the learning algorithms produce biased results.