3 May 2020 – Sunday – #49

It’s all about R.

Yesterday Spain let us out of our houses to exercise. It was a glorious jailbreak. First time in 1-1/2 months I got over 10,000 steps in a day. I didn’t quite follow the guidelines, but I kept my distance, didn’t touch anything except for my mobile and a pizza box, and washed my hands when I returned home.

The evening before I started isolation 50 days ago, I walked to Sagrada Familia. Last night I returned. They still haven’t finished construction.

Sagrada Familia, 2 May 2020

I probably should have returned home, but I was on a hunt for pizza. Two months without pizza is a lifetime. Somehow I ended up here.

Arc de Triomf, 2 May 2020

It was nice to see many people out walking! And scary. I headed back up Passeig de Sant Joan and found a pizza place. The pizza wasn’t perfect, but the weather was. Pizza and wine on the terrace last night. Hooray!

Sure, I was out a little too long, but yesterday was the first time people went outside in large numbers, so infection rates had to have been lower than they’ve been in months. They may be a little higher today because we went outside yesterday. As long as we keep R < 1, though, we can continue to go outside.

Barcelona and Spain have started the R dance after 1-1/2 months of lockdown.

Germany started its R dance a week ago. The naysayers already claimed it’s a failure based on a recent uptick in Germany’s R. German scientist point out, though, that relevant R readings won’t be available for another week. Even Germany, which has one of the best Covid-19 testing programs in the world, needs 10-14 days to measure which way R is headed after a change in Covid-19 restrictions.

Infections don’t show up immediately. Most people aren’t tested for the coronavirus until they start to show symptoms, which can take days to appear. Tests take time to process, and there is a further delay before positive results are reported to the authorities

The Telegraph, “No evidence of a second wave in Germany after lockdown lifted,” 1 May 2020

With R hovering around 0.75, Germany has some wiggle room. Every four people who get infected with Covid-19 infect three other people. Infections are waning. Chancellor Merkel and her government will decide whether to relax Covid-19 restrictions more next week when they have the relevant R measurements.

The R news isn’t so good on the Japanese island of Hokkaido, the prefecture hardest hit by Covid-19. The government instituted a model Covid-19 response with testing and a contact tracking program. However, anxious to get business started again and seeing Covid-19 cases drop significantly, the island lifted Covid-19 its restrictions on 19 March.

With hindsight, experts agree, it was too early and, just 26 days later and after 135 new infections were reported in the space of a week, the lockdown was reimposed on Hokkaido’s 5.3 million residents.

The Telegraph, “Japanese island suffering second wave of coronavirus after lifting lockdown too early,” 29 April 2020

Hokkaido lifted restrictions too soon. Its Covid-19 cases shot up again and it had to re-institute a full lockdown. The country of Japan is expected to extend its Covid-19 restrictions for another month, through 6 June.

I was talking to Cristián last night about applying Machine Learning to understand R, the rate at which the Covid-19 infection is spreading. Where testing data is weak, which is almost everywhere right now, the most reliable data on Covid-19 infections and R are hospitalizations and deaths. Unfortunately, those signals come 1-3 weeks after the actual infection, so R is detected retrospectively.

I hope someone is training Machine Learning models on non-testing data that could predict hospitalizations and deaths. Such Machine Learning models would provide an implied R from signals other than Covid-19 tests. Cell phone movement data, for instance, certainly seem like they would help predict R. We see less cell phone movement while people are sheltering in place.

What about using sewer system data? WHO has used polio detection in sewage to assess the effectiveness of its polio vaccine programs. The Michigan State Universiy has measured Hepatitis A in Detroit sewage, showing that the virus in waste water predicts viral outbreaks in the city about a week in advance. It’s no surprise, then, that researchers measuring Covid-19 in sewage see viral load increase about five days before an outbreak.

Researchers haven’t been able to predict Covid-19 case load, yet, but future work may provide a quick way to understand whether a Covid-19 lockdown is working and whether it’s safe to go out. Especially if sewage data is combined with other data in a Machine Learning model. For now, waste water analysis may be one of the best predictors of an impending Covid-19 outbreak and signal the need to lockdown at the city level or even by neighborhood.

As their testing becomes more reliable and precise, Dr. Medema and other researchers hope to zoom in on future outbreaks. Instead of looking at a wastewater treatment plant that handles an entire city or county, they may go down into manholes to monitor changes in individual neighborhoods.

New York Times, “Is It Safe to Come Out of Lockdown? Check the Sewer,” 1 May 2020

People are, in fact, training Machine Learning models to predict Covid-19 infections. It’s difficult to dig up news on how they are modeling R, though, because “R” is also a programming language.

The worst case R scenario? It’s shelter-in-place at a jail. Unfortunately prisons have been allowed to turn into Covid-19 herd immunity experiments. It seems as immoral to me as the Tuskegee syphilis experiments. Or providing unsafe work conditions for low wage workers in meatpacking facilities in order to keep the meat supply chain operating.

It appears 70% of the 150,000 inmates in the US federal prison system have contracted Covid-19. Federal prison officials claim they are doing everything possible, but don’t have guidelines for containing an epidemic in confinement. Meantime, prisoners are going to the hospital for Covid-19 and dying without any notification to next-of-kin. The prison guard labor union says prison guards are getting infected, too, which means the prisons function as Covid-19 hot spots infecting the communities around them. The prision systems are not only immoral, they’re stupid public health.

I’m going to finish today with a fun Twitter account Brad found.

Brad claims he broke the Bookcase Credibility story before the NY Times. I’m breaking it after everyone else. It’s that good. I hope they don’t offend your favorite pundit’s bookcase. Or maybe I do.

We’re out of our home prisons in Barcelona. For now. The R dance goes on.

%d bloggers like this: