Covid-19 : Brazil ICU,celebrating the latest Analytical innovations of Solidarité Covid- Fran?ais de Chine 1/3 Brazil

Since end of January 2020 with www.covidminute.com; we have collected an impressive amount of data and expertise on Covid pandemic dynamics. Some may think that all the brains of the world, the media, the scientists and other institutions who are delivering continuously smart analysis on the topic all over the press and Internet have offset our expertise, but those who think so are not those who follow the daily analytics on CovidFlow, the various analytics offered in the charts and the Greater China report delivered to 1200 expats in China on wechat groups. All these analytics feature some insights that are unique to the data sets we have collected and the curation we are doing on the data sets.

Here’s an overview in 3 articles of the latest analytical innovations we delivered.

1/3 Brazil Hospital Data set to track ICU occupancy and other metrics

Brazil is the 2nd largest country in cases and deaths. The more cases and the larger the country, the harder it is for a struggling country to monitor and report the metrics in interactive dashboards. Brazil Health Ministry reports every day the cases and deaths by region, yet for other key metrics (ICU occupancy, profile of patients), no aggregated data. 

Brazil Health authorities have yet implemented since the 2009 H1N1 epidemic a patient record tracker of all Acute Respiratory Syndrome Hospitalized patients which they have adjusted for Covid. This means we can analyze at patient level all the metrics we cannot pick up on the aggregated report. The task is not easy though considering the volume of the data set and I called on for help to ARTEFACT, Professionals of Big Data who kindly gave their support as a courtesy to analyze on PYTHON the whole data set of over 670K rows as of Sept 3 (of which 350K Covid +) and 134 columns. So we are very proud that Solidarité Covid - Fran?ais de Chine https://deeperin19coviddata.wordpress.com/, which operates only with volunteers with no publicity, very limited sponsors and donations of readers, is the first to publish analytics on this data set. The most relevant is the ICU occupancy history which started declining end of July, and hospitalizations starting to plateau 3 weeks later, with a very long plateau in nb daily deaths.

The program needs to run on all the back data since patients stay in ICU sometimes for weeks and takes about an hour to run at this stage. The notifications get updated retroactively so at each weekly update of the Open source file, the back data of ICU occupancy shows variance of about 1-9% in the last 4 weeks. One could argue that these updates may suggest that the drop of patients in ICU is not real, yet the gap from one update to the other would not exceed 9% and the drop in ICU we currently note is of 17% vs peak, and we are 4 weeks away from peak, so we are confident the drop is real and there will be light at the end of the tunnel for Brazil as well.

We have also looked at ages and co-morbidities of cases and deaths, but since we only have in the data set the hospitalized patients, only 9.1% of Brazil cases cumulated, the most severe, we are getting high lethality across all the comorbidities and age groups compared to other countries, including those like France where we do get metrics on Hospitalized patients (41% of hospitalized cases cumulated, 12% since May 13 and enhancement of testing capacity). We are also getting in Brazil low reporting rate on the comorbidities (44% of records with no entry on any of the co-morbidities, 32% of the deaths) so we don’t know whether those patients have no underlying condition or the fields were just skipped. In Europe countries we have in our benchmark, the co-morbidities are always reported on a sample of patients as to make sure they are accurately reported. We still believe the lethality in Brazil has been much higher, especially in the poorer areas (like Amazonia or shanty towns of Sao Paolo / Rio de Janeiro), where the access to medical care is poor.

Comparably with the Brazil Amazonian native population, The Navajo native American Indian nation in the USA (Arizona) has paid also a very high toll to the pandemic (504 deaths for a population of 173 K) with the highest US territory rate of deaths 1 M/Population (@ 2915).

The update of our Sociology of cases split by age, collected on respective websites of each country and regularly updated, shows Brazil Hospital with the highest share of younger age groups (and especially the 60-70 Years old) in deaths and highest lethality on 70 years +.

We are very proud of this achievement. Of course we'd be happy to share this data set with established institutions who are tracking Covid metrics through a partnership. In the mean time our readers can check the ICU Brazil data on CovidFlow, the interactive web pages on 215 countries and 50 USA States and on the charts.

https://deeperin19coviddata.wordpress.com

To be continued...