Search for Hidden Patterns in the Study of Coronavirus Patients Using Data Mining Methods

At the present time, when predicting the spread of the COVID-19 virus, the condition of patients with COVID-19 and the development of an antiviral vaccine, the method of artificial data analysis is actively used. In modern medicine, to improve the quality of medical care, increase the accuracy of diagnostics and reduce potential errors, methods based on artificial intelligence and machine learning are actively used. In this paper the reasons that affect the accuracy of the definition of coronavirus have been investigated. The set of patients is divided into five non-overlapping subsets (classes) according to the severity of the disease. To detect hidden patterns in the analysis, non-linear data transformations are used based on the calculation of the values of the class membership function for each feature, also used the formation of latent features based on generalized estimates. The values of thresholds between classes on the numerical axis are determined, both for individual features and generalized estimates of objects on the defined sets of features.