COVID-19 Trajectories monitoring
This article describes an online tool that I created. You can follow this link to interact with it; it is updated every day with new information.
Last weekend I created a tool that tracks every day the trajectory of the total confirmed cases and of COVID-19 by country. The tool is connected to the most reliable and most frequently updated data source of confirmed COVID-19 cases and deaths by countries.
The main indicator that is tracked is the speed or rate at which cases/deaths are increasing. Visually this can be seen by plotting the total cases vs the days since the first case was detected (a steeper line would indicate that cases are doubling at a faster rate and a flatter line would indicate slower rates of growth). Numerically, this can also be calculated with a compounded growth rate for a given period (e.g. for the last 7 days). This is a screenshot of some selected countries and their reported cases as of yesterday (March 26th):
Since my initial post in Facebook and LinkedIn, I have been improving the tool to track the countries’ compounded growth rates with the idea to be able to forecast which nations will be following the footsteps of countries such as China and South Korea, which seems to have a flattened curve and reached minuscule growth rates both for new reported cases and deaths.
To better understand the path that the countries follow, I created an animated scatter plot that tracks the rolling 7-day growth rate percentage for cases and deaths for every day of the year (see technical note at the end for the calculation). In the gif below, you can see what has been the speed of change of cases and deaths for all the countries.
As you can see in the animation, by the end of January, China had a growth rate above 40% for cases and above 30% for deaths. This was the worst moment of the epidemic inside the country (Hubei was locked down in January 23rd). Two weeks later, by the second week of February, the growth rate for cases was below 10%.
South Korea had a similar behavior: by day 58 of the year (Feb 27), the growth rates for cases and deaths were both above 40%. Two weeks later (day 72 of year), after the control measures were imposed by the government, both growth rates were around single digits.
It took Iran a bit longer to decrease the growth rate of cases to levels below 10%: 25 days (day 84 of the year, which is Tuesday March 24).
As the year progresses and affects other countries, we can see the different trajectories of the growth rates. Italy had the worst rates of change during the last week of February (>50% for cases and >40% in deaths) and imposed total national lockdown by March 9. As of today (17 days after the national lockdown), the growth rates for cases is 10.1% and for deaths 13.4%
Spain enacted lockdowns in March 14th and (12 days later) they still have a 18% growth rates for cases and 26% for deaths. Even though Spain has not reached the single digit growth rates, you can follow their evolution in the animation which has a slowing down trend.
Other countries like the US, which have not yet actively advised or imposed self-isolation measures, are experiencing positive growth rates (acceleration of new cases and deaths instead of slowing down the pace). The latest data point shows gowth rates for both cases and deaths in the range of 30%.
What can we learn from this?
Even though every country has a unique set of conditions that reflects its own circumstances, we could try to generalise some learnigs from the animation:
- Most countries have a “counterclockwise” movement, where initially there is a high growth rate in the cases; in a second moment the growth rate of cases is constant while the growth rate of deaths increases. Finally, both growth rates tend to decrease once the governments have imposed some kind of control to the spread.
- The growth rate of cases ussually decreases much faster than the growth rate of deaths. To see my point, take a look at the same animation, where I have highlighted 3 countries of Asia.
- The lockdown measures work. See the same animation for Italy and Spain, the worst-hit european countries that have imposed very restricted measures of social-distancing in days 69 and 74 respectively
- As the year progresses, we can see how the countries in different continents have progressed through the phases described earlier. February saw high rates of change in Asia and some countries in Europe. The first weeks of March saw the same period for many other counries in Europe and the US and Australia. As we come close to the end of March, most countries in Asia are achieving low growth rates in cases and deaths; at the same time more countries in Africa, North and South America are entering the phase of accelerated growth. See this side by side animation to watch the trend:
As a rule of thumb, all the countries that have growth rates that exceed the 30%+ mark need to do a better job at containment of the situation. The direction of the trend is also crucial, the closest the countries manage to get to single digit growth rates, the faster they will emerge from this crisis.
This siuation is evolving very quickly and the information shown here will be obsolete tomorrow. To access current figures, the tool that I described at the beginning of the blog will update automatcally every day with the latest information released by the John Hopkins University. In the section below the main graphs, you can find the snapshot of the current date of the rate of change by country:
If you want to connect and have a conversation about this or any other topic, please feel free to find me on LinkedIn:
Just a small technical note: the compunded growth rate you see is calculated for every day of te dataset, against the figure that the country had 7 days before. For example, the calculation for cases would be (# cases current day/# cases 7 days ago)^(1/7) where 7 the periods elapsed — my growth window. I choose a rolling figure to track the rate of change as frequently as possible, as opposed to group by weeks.