What is it about?
The world has been in the throes of Covid-19 since 2020, with countless infections and millions of deaths worldwide. Effectively managing a pandemic necessitates tracking of infections and predictive modelling of its behaviour so that the spread can be contained in time. To this end, the author of this new study published in Statistical Communications in Infectious Diseases applied the generalised linear model (GLM) approach involving the autoregressive process with log link to model new cases and new deaths occurring daily due to Covid-19 in Turkey. Government data between March 11 and September 15, 2020, were used to train the model. Predictions made using the model fit the data well for daily new cases, new deaths, and the effective reproduction number i.e., the average risk of secondary infection per infectious case, as suggested by diagnostic plots and goodness of fit statistics. In the article, the author also reviews the mathematical approaches adopted in similar studies conducted in Turkey and other countries. The investigation reveals that many of these models do not consider parameters such as interventions, outliers, and the over dispersion problem. Additionally, the author also highlights the importance of curfews in curbing new cases of Covid-19.
Photo by Sarah Kilian on Unsplash
Why is it important?
Studies modelling Covid-19 behaviour are fundamental for effective disease management alongside the medical studies looking into cures and vaccines. Authorities need statistical tools that model disease behaviour accurately to institute public regulations, plan vaccination drives, and formulate policies to curb infection spread. This study offers a simple yet effective tool that makes use of well-established statistical methods to estimate the number of daily new cases and new deaths while considering often ignored parameters like over dispersion and data anomalies. KEY TAKEAWAY: The GLM approach involving autoregression with log link provides consistent and robust estimates of Covid-19 disease behaviour in Turkey and may be applied in other countries.
Read the Original
This page is a summary of: GLM based auto-regressive process to model Covid-19 pandemic in Turkey, Statistical Communications in Infectious Diseases, January 2021, De Gruyter, DOI: 10.1515/scid-2020-0006.
You can read the full text:
Predicting the spread of COVID-19 infection accurately
For any model, the estimations are only as good as the data used. The cases reported during the early stages of the pandemic are critical. The disease spread should be modeled at different stages of the pandemic. Taking input variables specific to each region into account can also improve estimations.
COVID-19 infection curves and measures to curb its spread
Understanding infection trends and identifying measures that work are key to applying the best strategies to bring the pandemic to an end as fast as possible.
Be the first to contribute to this page