Estimates we made

An archive of all estimates we made during the growing phase and peak.

Confirmed represents the cumulative number of confirmed cases in the country as per the country's health authority, or by WHO. It is updated daily.

Estimated highest and lowest limits provide estimated cumulative number of confirmed cases in the next few days for the country, it is based on an exponential regression of the country's values known at the day of the estimate. Because the country being assessed has a small number of data points, this line is not a good estimate, we use it as a reference to assess errors.

Spain and Italy were preselected based on the correlation of its values with Portugal's, being countries with similar social behaviour, are at least 4 days ahead in the epidemy, and considering the measures taken by each country.

All exponential regression parameters are free allowing us to monitor which model is representing better the actuals. During the first days the exponential regression was used. The margin of error, in a way, may be considered by the exponential’s multiplier which we consider that at t=0 should be between 2 and 4.

The estimates and projected data points are based on a very simple concept: the virus will behave in Portugal as it did in Italy or Spain, because those are countries with similar social behaviours as Portugal and have a few more corelated factors.

There are highly sophisticated epidemy models as is SEIR (Susceptible-Exposed-Infected-Recovered compartmental model) amongst others, which depend on a variety of variables. For this new Coronavirus there isn’t enough data and many uncertainties, therefore sophisticated models will not have a statistically high level of confidence, that being the reason why we may be seeing governments and experts giving very different numbers or changing measures and actions within one or two days. Some of SEIR’s parameters for which there is lack of certain data include:

  • Parameters for the reproductive number and incubation period of the virus
  • duration of the recovery period
  • Transmission of infection from asymptomatic individuals during the incubation period
  • Transmission of infection from people that already recovered and infected again
  • Mobility of residents within each region, which varies from country to country and region to region
  • The number of people tested
  • The change of behaviour of the population (risk tolerance) to this type of situation, varies between different cultures, GDP, experience with similar outbreaks.
  • Demographics
  • Social behaviours, family size, etc…
  • Updated Travel data (airport to airport)

Therefore, in March 2020 it is difficult to establish an accurate sophisticated model to predict the epidemic.

We believe our model, using the data of countries ahead in the infection and with similar social behaviours, can predict 2-week trends, updated and adjusted weekly, and provide vital information to all.

Accurate Information and estimates allow decision making. In the next few months, health authorities, governments, hospitals will take decisions on how to care for the maximum amount of people with the limited resources at their disposal.

As stated in “How are we modelling the projected values?”, at this moment there isn’t enough data and many uncertainties to use a sophisticated model, therefore, for evaluating one may want to develop expedite methods which can be quickly adjusted to what is happening. We expect our method can estimate reasonable values in an easy and rapid way.

As generals establishing strategies and tactics, in this war it is mandatory to manage the limited amount of doctors, nurses, other hospital staff, beds in ICUs, beds in negative pressure rooms, ventilators, transports, medication distribution, etc… and to manage each day considering what will be needed for the whole war. Wars are won with logistics and resource management.

Therefore, better information and quickly adjustable projections to the conditions and actions taken allow for better management of resources, hopefully allowing a better result, in this case to help as many people as possible to see the end of the war.

Strategy without tactics is the slowest route to victory. Tactics without strategy is the noise before defeat

Sun Tzu

The existing information is that infected people that develop COVID-19 show symptoms up to 14 days after being infected. The measures and actions that Governments and Health Authorities take, and what each one of us does to protect themselves, will only have impact 10 to 12 days from the date that measures limiting social interaction were implemented and/or behaviours changed. This was validated by the quarantine measures implemented in China.

The specific cases of Portugal, Spain and Italy are being considered as the comparable data sets. Italy has imposed a forced quarantine in Lombardy on the 8 March and extended it to the whole country on 10 March, therefore only between 19 and 22 march the impact of such measures (if any) can be measured. Spain took different measures at different stages.

Portugal has taken some actions towards voluntary social isolation on the 13 March and additional measures on the 16 March, therefore only from 23 March we will be able to measure the impact of such actions.

Between 20 and 25 March we can assess individually and also compare Italian, Spanish and Portuguese and what impact (if any) the Governments’ actions had, assess the impact of Forced Quarantine versus Voluntary Self Isolations, etc…) and do new projections for the following 2 weeks.

Different actions will have different impacts in different societies and cultures. The individual behavior of each of us represents the measure of our respect and humanity for the health of the person living and working around us. It is essential to follow the recommendations of the Health Authorities and reduce the contagion rate.

We are constantly developing new charts that will appear over the next few days.