XLSTAT: COVID-19 Data Analysis in Excel

 

Setting Up

Select the Life Science menu/Survival Analysis/COVID-19 command.

Clicking the COVID-19 icon displays a dialog box with COVID-19 information.

1. Use "Active workbook" to import data already loaded, ideally from the first sheet with the same format as ECDC.

To get the latest data, choose "Import from ECDC," which updates automatically daily. You can also select your own data under "Select your data," ensuring you include cases, deaths, dates, and entities.

Use "Filter out countries" to exclude specific countries. Additional options allow focusing on particular countries for the model’s predictions from outbreak start to present.

Only one model is available: the double exponential Holt-Winters smoothing. You can select how many past days to include.

Interpreting the Results

Clicking OK prompts XLSTAT to analyze data, generate summary statistics, predictions, and charts. The first table shows stats for each unfiltered country. For outbreaks before January 1, 2020, data prior to that date is merged into a single dataset.

After the start date, you'll see data on days since the first case, total cases, total deaths, daily max cases/deaths, latest counts, next-day predictions, and doubling time for cases and deaths. A second table lists the 10 countries with the most recorded deaths.

A chart follows, showing the days until cases/deaths double (shown only if 10 or less or computable). It displays each country's trajectory since deaths exceeded 2, with a log scale on the y-axis for comparison.

A race chart (Excel 2010+) appears, showing the daily evolution of fatalities for the top 10 countries, with the date displayed at the top.

If a specific country is selected, the last table shows daily model predictions and estimates cases for the next 5 days, assuming no changes or worsening factors. The model may be less accurate after peak periods. For Italy on March 31, the last two columns estimate prediction error based on previous errors, providing a more realistic measure than confidence intervals.

....

The last chart displays the data and the model predictions for both cases and deaths.