Introduction: Medical statistics is one of the "milestones" of current medical systems. It is the foundation for many protocols, including medical care systems, government recommendations, epidemic planning, etc. At this time of global COVID-19, credible data on epidemic spread can help governments make better decisions. This study's aim is to evaluate the cyclicity in the number of daily diagnosed coronavirus patients, thus allowing governments to plan how to allocate their resources more effectively.
Methods: To assess this cycle, we consider the time series of the first and second differences in the number of registered patients in different countries. The spectral densities of the time series are calculated, and the frequencies and amplitudes of the maximum spectral peaks are estimated.
Results: It is shown that two types of cycles can be distinguished in the time series of the case numbers. Cyclical fluctuations of the first type are characterized by periods from 100 to 300 days. Cyclical fluctuations of the second type are characterized by a period of about seven days. For different countries, the phases of the seven-day fluctuations coincide. It is assumed that cyclical fluctuations of the second type are associated with the weekly cycle of population activity.
Conclusions: These characteristics of cyclical fluctuations in cases can be used to predict the incidence rate.
Keywords: Coronavirus; autoregression; dynamics; spectral analysis; time series; waves; week end effect.
© 2021 Elsevier Ltd. All rights reserved.