Understanding the cycles of COVID-19 incidence: Principal Component Analysis and interaction of biological and socio-economic factors

Abstract

The incidence curve of coronavirus disease 19 (COVID-19) shows cyclical patterns over time. We examine the cyclical properties of the incidence curves in various countries and use principal components analysis to shed light on the underlying dynamics that are common to all countries. We find that the cyclical series of 37 countries can be summarized in four principal components which explain over 90% of the variation. We also discuss the influence of complex interactions between biological viral natural history and socio-political reactions and measures adopted by different countries on the cyclical patterns exhibited by COVID-19 around the globe.

Anxiety in anesthesia providers during coronavirus disease 19 pandemic: Insights into perception of harm a cross-sectional study

“…On the other hand, since our data was collected and analyzed before the widespread use of vaccines, further research is required to answer the same question in this new context. Some authors have postulated that the viral incidence cycles on just a few factors [ 21 ], studies with longitudinal design are necessary to clarify the relationship between anxiety and the fluctuation of cases during a pandemic. The strengths of our study include the high response rate to the survey, the use of a well-validated tool for assessing anxiety, and the low likelihood of recall bias conferred by the prospective recollection of data over a short time period.…”

Differences in COVID-19 cyclicity and predictability among U.S. counties and states reflect the effectiveness of protective measures

“…The COVID-19 pandemic has stimulated the development of numerous quantitative models to help understand and forecast disease dynamics, and to assist public health decision-making (e.g. 12,22,43). Rather than develop methods for making predictions, in this study we have focused on the inherent unpredictability of COVID-19 dynamics.…”

“…Preprint posted on medRxiv, url: https://www.medrxiv.org/ Of the 100 counties and 49 states, 96 and 41 showed cyclic dynamics in the stationary domain (SI Appendix, Fig.S6), and all analyzed jurisdictions had mean r(t) of nearly zero (not shown). The estimated period was similar at the county and state levels (SI Appendix, Fig.S7a,b): counties had a median of 23 weeks (interquartile range[20][21][22][23][24][25][26][27][28][29], and states had a median of 26 weeks(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33). The damping factor (d) was also similar (SI Appendix, Fig.S7c,d): counties had median d = 0.91 (0.85-0.96), and states had median d = 0.91 (0.83-0.94).…”

“…An advantage of our approach is that we can associate predictability to specific dynamical patterns observed during the pandemic, like cyclic dynamics, which potentially lead to more accurate predictions (e.g. 21,22).…”

Application of principal component analysis on temporal evolution of COVID-19

“…The relations between the spread rates of COVID-19 in high risk countries are determined by first Two PCs [ 15 ]. The first four principal components of PCA is used to investigate the cyclical patterns of the COVID-19 in different countries [ 16 ]. PCA is applied on data of 213 COVID-19 patients and identified three distinct groups of the COVID-19 patients [ 17 ].…”