Data Science newsletter – April 27, 2020

GROUP CURATION: N/A

A comparison of daily deaths in Italy since January 2020 with those over the previous five years there indicates that the fatality rate in that country for those infected with the new coronavirus is at least 0.8%, far higher than that of the seasonal flu and higher than some recent estimates.

Extrapolating from the Italian data, University of California, Berkeley, and Lawrence Berkeley National Laboratory data scientists estimate that the fatality rate in New York City and Santa Clara County in California can be no less than 0.5%, or one of every 200 people infected.

These conclusions contrast with those of a study posted online last week by Stanford University epidemiologists, who pegged the fatality rate at between 0.1% and 0.2%. An affiliated team from the University of Southern California (USC) this week reported a similar fatality rate in Los Angeles.

“Their final number is much lower than our estimate,” said senior author Uros Seljak, a UC Berkeley professor of physics.

A civil liberties organization is raising the alarm after the province of Ontario began sharing COVID-19 test results with police and other organizations.

On April 6, Ontario officials announced that first responders in the province, including police, firefighters, and paramedics, would have access to a database that includes names, addresses, dates of birth, and whether a person had tested positive for COVID-19.

The Canadian Civil Liberties Association is calling it “an extraordinary invasion of privacy.”

Scientists involved with several ongoing Mars missions are warning that a budget proposal that would end one mission and curtail operations of others could disrupt NASA’s broader Mars exploration architecture.

NASA’s fiscal year 2021 budget request, released in February, proposed terminating the 2001 Mars Odyssey orbiter, which received $11.5 million in fiscal year 2019, and cutting funding for the Curiosity rover by more than 20% from 2019 levels to $40 million.

The proposed cuts came despite favorable reviews of both missions in the latest senior review of extended missions. “There’s a cloud hanging over our mission,” said Jeff Plaut, project scientist for Mars Odyssey, during an online meeting of NASA’s Mars Exploration Program Analysis Group (MEPAG) April 17, referring to the proposed cancellation. “This was despite the fact that senior review gave Odyssey a very favorable review and recommended actually an increase in our funding.”

The 2020 AI 2000 Annual List recognizes 200 Most Influential Scholar Award winners and 1800 nominees, 1126 of which are from the United States, 174 from China, and 307 from the European Union. Among the list, many of the scholars have made outstanding contributions to several fields, such as Yoshua Bengio and Alex J. Smola—each recognized for contributions to four distinct fields. The detailed statistics of the 2020 AI 2000 Most Influential Scholar List are provided below, including distributions of multi-field winners, countries, institutes, and gender.

Issues of censorship surrounding the publication of scholarly research in China have been prominent since a series of press reports and publisher statements revealed that works had been removed from circulation that were deemed sensitive by Chinese buyers. As George Cooper observes, evidence that Chinese authorities are conducting pre-publication vetting of COVID-19 related research, raises new challenges for publishers seeking to distribute open access research papers on this subject, as there is little ground for publishers to remove these papers from their platforms. As publisher commitments to openness collide with their obligations to operate within the legal frameworks of the countries they operate in, it is argued that COVID-19 presages an overdue discussion on the limits of openness in publishing.

Nearly half the “people” talking about the coronavirus pandemic on Twitter are not actually people, but bots, according to new research from Carnegie Mellon University.

And many of those bots are rapidly feeding Twitter with harmful, false story lines about the pandemic, including some inspiring real-world activity, such as the theory that 5G towers cause COVID-19, or state-sponsored propaganda from Russia and China that falsely claims the U.S. developed the coronavirus as a bioweapon or that American politicians are issuing “mandatory” lockdowns.

What is AQ?
You split population to two halves. Then work in weekly shifts. One week Group 1 goes to work & Group 2 stays home. The next week, they switch roles.
Our paper shows that it is probably as good as a COMPLETE 80% quarantine, despite allowing 50% activity at all times [thread]

Nearly two months after an embarrassing end to his presidential campaign, Mike Bloomberg is again deploying his massive personal fortune – this time to combat the coronavirus.

The billionaire former New York City mayor is spending tens of millions of dollars to bolster social services, feed first responders and help local officials trace the spread of the coronavirus in the city that has become the epicenter of the pandemic in the United States.

The effort is in line with Bloomberg’s long-standing interest in scientific research and public health. But it’s also an opportunity for him to rebuild an image that was badly damaged earlier this year by accusations from some Democrats that he was trying to buy the White House. He spent more than $1 billion on a campaign that ultimately resulted in just one primary victory, in American Samoa.

The initial coronavirus outbreaks on the East and West Coasts emerged at roughly the same time. But the danger was communicated very differently.

Dozens of hospitals across the country are using an artificial intelligence system created by Epic, the big electronic health record vendor, to predict which Covid-19 patients will become critically ill, even as many are struggling to validate the tool’s effectiveness on those with the new disease.

The rapid uptake of Epic’s deterioration index is a sign of the challenges imposed by the pandemic: Normally hospitals would take time to test the tool on hundreds of patients, refine the algorithm underlying it, and then adjust care practices to implement it in their clinics.

“Climate change is the most significant crisis facing mankind right now,” Professor Paul Barford, of the University of Wisconsin, told DCD. And, as part of that, the fabric of the Internet also faces the consequences of a warming planet.

With this in mind, a few years ago Barford and a team of researchers set out to answer a relatively straightforward question: “What is the risk to telecommunications infrastructure, of the sea level rise that’s projected over the next hundred years?”

Our bipartisan group of experts in economics, public health, technology, and ethics from across the country, with support from The Rockefeller Foundation, has released the nation’s first comprehensive operational roadmap for mobilizing and reopening the U.S. economy in the midst of the COVID-19 crisis.

“Roadmap to Pandemic Resilience: Massive Scale Testing, Tracing, and Supported Isolation (TTSI) as the Path to Pandemic Resilience for a Free Society,” lays out how a massive scale-up of testing, paired with contact tracing and supported isolation, can rebuild trust in our personal safety and re-mobilize the U.S. economy.

Among the report’s top recommendations is the need to deliver at least 5 million tests per day by early June to help ensure a safe social opening. This number will need to increase to 20 million tests per day by mid-summer to fully re-mobilize the economy.

Google and Apple’s COVID-19 contact tracing technology is getting improved ahead of launch, with some notable improvements to protecting privacy.

The tool, which will roll out to developers in beta next week, will allow users to see when they have come within close contact with someone who has tested positive for COVID-19. It will work across both iOS and Android devices.

Countries are rushing to develop apps to give a detailed picture of the risk of catching the coronavirus, as the chain of infection is proving hard to break because it can be spread by those showing no symptoms.

Chancellery Minister Helge Braun and Health Minister Jens Spahn said in a joint statement that Berlin would adopt a “decentralised” approach to digital contact tracing, thus abandoning a home-grown alternative that would have given health authorities central control over tracing data.

In Europe, most countries have chosen short-range Bluetooth “handshakes” between mobile devices as the best way of registering a potential contact, even though it does not provide location data.

The Centers for Disease Control and Prevention (CDC) plans to roll out an app in May that will accelerate electronic case reporting of COVID-19 cases.

The app, based on Fast Healthcare Interoperability Resources (FHIR) standards, can be implemented quickly to automate COVID case reporting, said Laura Conn, health scientist and eCR lead in the CDC’s Center for Surveillance, Epidemiology, and Laboratory Services.

The app enables healthcare providers that don’t have the capability to automatically send case reports from their health IT systems to more efficiently send data to public health agencies.

“The FHIR app can be rapidly implemented to automate COVID-19 electronic case reporting (eCR) in otherwise non-enabled EHRs. The app will connect COVID-19 eCR to existing infrastructure to confirm cases and route to appropriate public health surveillance systems. We are making sure that can be easily used without requiring a software release by the vendors,” said Conn during a meeting this week of a Congressional health IT advisory committee (HITAC).

Online April 26-30. “We have released a call for virtual rooms and socials that makes it possible for everyone to more easily meet.” [$$$]

Online April 26-30. “Here’s a sneak peek at a part of the conference portal: switch your watching next week and scroll for papers, videos and discussion instead.” [$$$]

“We propose SPECTER, a new method to generate document-level embedding of scientific documents based on pretraining a Transformer language model on a powerful signal of document-level relatedness: the citation graph. Unlike existing pretrained language models, SPECTER can be easily applied to downstream applications without task-specific fine-tuning. Additionally, to encourage further research on document-level models, we introduce SciDocs, a new evaluation benchmark consisting of seven document-level tasks ranging from citation prediction, to document classification and recommendation. We show that SPECTER outperforms a variety of competitive baselines on the benchmark.”

In this talk from the Machine Learning Conference, Kamil Kaczmarek and Jakub Czakon focus on practical guidelines and tips on how to set-up and maintain smooth collaboration in data science projects. Discover how to best track and collaborate in data science. [video, 31:25]

“As part of the broader PyTorch community, Facebook AI and AWS engineers have partnered to develop new libraries targeted at large-scale elastic and fault-tolerant model training and high-performance PyTorch model deployment. These libraries enable the community to efficiently productionize AI models at scale and push the state of the art on model exploration as model architectures continue to increase in size and complexity. Today, we are sharing new details on these features.”

“The interpretability of neural networks is becoming increasingly necessary, as deep learning is being adopted in settings where accurate and justifiable predictions are required. These applications range from finance to medical imaging. However, deep neural networks are notorious for a lack of justification. Explainable AI (XAI) attempts to bridge this divide between accuracy and interpretability, but as we explain below, XAI justifies decisions without interpreting the model directly.”

“This post will explore some of the tools and platforms that can help with a key stage of the online research process: creating your survey or experiment. Specifically, we’ll be looking at options for running online experiments, with a slight focus on the more complex platforms – those designed to collect reaction time data (e.g., cognitive tasks), or to deliver complex experimental paradigms with a range of response types.”