None of the studies have yet published peer-reviewed results, but we’re getting the first evidence that the idea works. On Thursday, researchers at WVU’s Rockefeller Neuroscience Institute reported that Oura ring data, combined with an app to measure cognition and other symptoms, can predict up to three days in advance when people will register a fever, coughing or shortness of breath. It can even predict someone’s exact temperature, like a weather forecast for the body.
Professor Ali Rezai, the institute’s director, said the technology is valuable because it’s tuned to reveal infection early on, when patients are highly contagious but don’t know it. He calls the combination of the smart ring and app a kind of “digital PPE,” or personal protective equipment. ‘It can say, “This individual needs to stay home and not come in and infect others,’” he said.
There’s more: Researchers at Stanford University studying changes in heart rate from Fitbits tell me they’ve been able to detect the coronavirus before or at the time of diagnosis in 11 of 14 confirmed patients they’ve studied. In this initial analysis, they could see one patient’s heart rate jump nine days before the person reported symptoms. In other cases, they only saw evidence of infection in the data when patients noticed symptoms themselves.
“The bottom line is it is working, but it’s not perfect,” said Stanford professor Michael Snyder.
“I don’t think these tests are ready for clinical use yet,” said University of California-San Francisco immunologist Dr. Alexander Marson, who has studied their reliability. He and his team vetted 12 different antibody tests and found all but one turned up false positives — implying that someone had antibodies when they didn’t ― with false-positive rates reaching as high as 16%. (The study is preliminary and has not been peer-reviewed yet.)
More than 100 antibody tests are currently available in the U.S., including offerings by commercial labs, academic centers and small entrepreneurial ventures. As serious questions emerged earlier this month about the accuracy of the tests and the usefulness of the results, the U.S. Food and Drug Administration said it will require companies to submit validation data on their products and apply for emergency-use authorizations for their products.
… The current coronavirus outbreak provides a few examples of detect-to-protect technologies that have helped minimize damage. The pulse oximeter—a device worn on the finger that measures blood oxygenation in patients—has been promoted(5) as a vital early warning tool in dealing with the puzzling problem of “happy hypoxics”, coronavirus-infected patients who feel and appear fine, but have critically low levels of oxygen in their blood.(6) Other examples include the airborne particle counters that come with many home HEPA air purifiers (my wife calls it a dog detector because it flips on high every time our furry dog walks into the bedroom), the infrared cameras used to measure body temperature of passengers walking through airport terminals, and kits containing nontoxic fluorescent dyes and ultraviolet flashlights being sold as a visual aid to teach people better handwashing protocols (see glogerm.com).
There are scientific and commercial challenges facing emerging detect-to-protect technologies: the science side involves identifying the sensing problem and its best solutions, while the commercial side involves identifying the paths to translating the most promising concepts into the real world. Some technologies might be excellent detect-to-protect solutions for problems that are far removed from what their inventors had in mind. Yet others will remain of dubious value forever. This challenge is made more difficult by the lack of a “killer application” for translation of many low-fidelity detect-to-protect sensing systems. Even very high-fidelity detect-to-treat sensing systems face this challenge when the small problem they solve just does not have a sufficiently broad market. The current pandemic underscores this issue in a stark and painful way.
In the two years since it set up shop, insitro has been plugging away at a data- and machine learning-driven approach to drug discovery. Now, the company is picking up $143 million to build out its technology, pursue new targets and advance treatments for genetically defined patient groups.
Insitro is using in vitro systems—or “test-tube experiments” involving cells outside the human body—to predict what drug developers would see in a human, in vivo, clinical model. These models could propose new drug targets and predict how patients, or subgroups of patients will respond to specific treatments.
With doctors and researchers around the world searching for effective treatments for COVID-19, many drugs approved to treat other diseases are being used in hopes that they’ll be effective against the virus, a use that’s known as “off-label.” New research from the Perelman School of Medicine at the University of Pennsylvania catalogued every use documented in medical literature so far and found physicians have reported on the use of more than 100 different off-label and experimental treatments. The effort, called COvid19 Registry of Off-label & New Agents (CORONA), is an attempt to take an inventory of what’s being used where, as well as to spot any evidence of treatments that warrant further investigation in a randomized clinical trial. The findings published in Infectious Diseases and Therapy today.
“We can’t win this fight if we don’t take stock of the tools that are already being used and search for new ones that could be effective. While off-label use is happening all over the world, there’s currently no system in place to track it, so we felt like we had to create one,” said the study’s lead author David C. Fajgenbaum, MD, MBA, MSc, an assistant professor of Translational Medicine & Human Genetics and director of the Center for Cytokine Storm Treatment & Laboratory (CSTL) at Penn.
Vaccines must be safe and effective. Once (and only once) this is proven, immunization campaigns need to be comprehensive to succeed. But this presents many challenges. For low-income countries, and in those without universal health care, a key obstacle is ensuring that vaccines are available and affordable. For certain higher-income countries — for example some in Europe — the challenge for coronavirus will be to overcome scepticism about vaccines, which is being fuelled by false information.
Researchers can play a part. Knowing what to do in the middle of a pandemic isn’t straightforward. But for those considering how to respond to the kinds of questions that everyone is asking, and what to do about disinformation, there are ways to help.
The study lacked two kinds of clearance, the FDA said: a federal emergency use authorization, a type of pandemic-era green light used to speedily clear tests and medical devices during an emergency, and approval from an outside group of experts tasked with providing ethical oversight for research, known as an institutional review board. Obtaining either clearance could have allowed the effort to go forward, according to the FDA.
Outside experts decried FDA’s decision, saying the abrupt move was unnecessary, confusing, and counterproductive. Rather than shutting down the SCAN effort, the agency could have asked them to re-apply for approval from the ethics board, for example.
Some studies are beginning to indicate that when patients are severely ill, the virus is replicating deeper in the respiratory system, beyond the reach of the swabs used for most of the testing, Misialek said.
“The traditional way that we diagnose virus is usually through a nasopharyngeal swab that goes up into the back of the nasal passage,” he said. But what if the virus isn’t replicating there? A swab test will indicate someone is virus-free, even as the coronavirus is busy replicating in the lungs or even in the intestines. “As the disease progresses, the likelihood of a positive result goes down,” Misialek said.
Other factors affect testing accuracy, also. “It can depend on the stage of disease,” Misialek said. “Virus is most likely to be detected one to two days before they are actually symptomatic, and up to four days after. That’s about a one-week window you have to catch it.”
Even if people do have phones, we know that people do not necessarily use their technology similarly. The current automatic contact tracing apps assume technology maps 1:1 with people. I’ve seen in my own work where participants share smartphones depending on which family member needs the most connectivity within a certain time period. Thus, to be effective in these situations, a contact tracing app would have to log who the user is during the contact time.
Similarly, there is an assumption that participants have their Bluetooth on all the time. In our work, participants often reported turning Bluetooth and wifi off to save battery power. This was especially true where participants were working long days and needed their phone to have enough battery to connect with loved ones (e.g., a cashier at the grocery store; facilities staff at a hospital).
arXiv, Physics > Physics and Society; Kyle R. Myers, Wei Yang Tham, Yian Yin, Nina Cohodes, Jerry G. Thursby, Marie C. Thursby, Peter E. Schiffer, Joseph T. Walsh, Karim R. Lakhani, Dashun Wang
from
The COVID-19 pandemic has undoubtedly disrupted the scientific enterprise, but we lack empirical evidence on the nature and magnitude of these disruptions. Here we report the results of a survey of approximately 4,500 Principal Investigators (PIs) at U.S.- and Europe-based research institutions. Distributed in mid-April 2020, the survey solicited information about how scientists’ work changed from the onset of the pandemic, how their research output might be affected in the near future, and a wide range of individuals’ characteristics. Scientists report a sharp decline in time spent on research on average, but there is substantial heterogeneity with a significant share reporting no change or even increases. Some of this heterogeneity is due to field-specific differences, with laboratory-based fields being the most negatively affected, and some is due to gender, with female scientists reporting larger declines. However, among the individuals’ characteristics examined, the largest disruptions are connected to a usually unobserved dimension: childcare. Reporting a young dependent is associated with declines similar in magnitude to those reported by the laboratory-based fields and can account for a significant fraction of gender differences. Amidst scarce evidence about the role of parenting in scientists’ work, these results highlight the fundamental and heterogeneous ways this pandemic is affecting the scientific workforce, and may have broad relevance for shaping responses to the pandemic’s effect on science and beyond.
In my opinion peer review is far from perfect. But with determination new and revolutionary ideas can get through the peer review process, though it may take some years. The problem is, of course, that most revolutionary ideas are wrong, so peer review tends to stomp hard on all of them. The alternative is to have everyone self publish and that is what is happening with the arXiv distribution service. Papers are getting posted there with no intent of ever undergoing peer review, and so they are effectively getting published with no review. This can be seen as part of the problem of populism where all self proclaimed experts are listened to with equal authority, and so there is no longer any expertise.
In this episode of the McKinsey on AI podcast mini-series, we share a conversation between James Manyika, co-chairman of the McKinsey Global Institute, and University of California, Berkeley, professor Stuart Russell. They explore how we can ensure AI truly benefits humanity rather than causing us harm. According to Russell, doing so begins with abandoning the idea of creating “intelligent” machines altogether. [audio, 20:37, and transcript]
If you think fact-checking is hard, which it is, then what would you say about verifying scientific claims, on COVID-19 no less? Hint: it’s also hard — different in some ways, similar in some others.
Fact or Fiction: Verifying Scientific Claims is the title of a research paper published on pre-print server Arxiv by a team of researchers from the Allen Institute for Artificial Intelligence (AI2), with data and code available on GitHub. ZDNet connected with David Wadden, lead author of the paper and a visiting researcher at AI2, to discuss the rationale, details, and directions for this work.
bioRxiv; Tom Hope, Jason Portenoy, Kishore Vasan, Jonathan Borchardt, Eric Horvitz, Daniel S. Weld, Marti A. Hearst, Jevin West
from
The COVID-19 pandemic has sparked unprecedented mobilization of scientists, already generating thousands of new papers that join a litany of previous biomedical work in related areas. This deluge of information makes it hard for researchers to keep track of their own field, let alone explore new directions. Standard search engines are designed primarily for targeted search and are not geared for discovery or making connections that are not obvious from reading individual papers.
In this paper, we present our ongoing work on SciSight, a novel framework for exploratory search of COVID-19 research. Based on formative interviews with scientists and a review of existing tools, we build and integrate two key capabilities: first, exploring interactions between biomedical facets (e.g., proteins, genes, drugs, diseases, patient characteristics); and second, discovering groups of researchers and how they are connected.
Online June 8-10. “For the first time ever, MIT Technology Review will partner with Harvard Business Review, the authority on leadership and management, to host sessions and speakers offering trusted guidance on navigating change in times of uncertainty.” [registration required]
Online June 4, starting at 3 p.m. BST. Speakers: Catherine D’Ignazio and Lauren Klein. “After the webinar, we will host Catherine and Lauren for virtual “after talk drinks”. This session will be 45 minutes and try to capture the conversations that we would have over a glass of wine or ginger beer to explore the themes of their presentation.” [registration required]
“NSF and Amazon are partnering to jointly support computational research focused on fairness in AI, with the goal of contributing to trustworthy AI systems that are readily accepted and deployed to tackle grand challenges facing society. Specific topics of interest include, but are not limited to transparency, explainability, accountability, potential adverse biases and effects, mitigation strategies, algorithmic advances, fairness objectives, validation of fairness, and advances in broad accessibility and utility.” Deadline for proposals is July 13.
APIs are the building blocks of today’s digital world: developers use them to quickly integrate features, data, services and functions into their own apps, removing the need to build and scale all those elements from scratch themselves. Given its DNA as a dev tool, an API-as-a-marketplace is a block in building a business just like Stripe (for payments), Twilio (for communication), Postmates (for on-demand delivery), etc.
But unlike Stripe, Twilio, and Postmates, the supply of an API-as-a-marketplace is not necessarily a commodity. That is, should they choose to, developers and businesses (or even their end users) can pick their suppliers based on whatever requirements they have.
For example, on Shippo, businesses may want to optimize the carrier (supplier) based on price, visibility on tracking, etc. – after all, as CEO Laura Behrens Wu says “Shipping is not one size fits all”. On Patch, businesses can choose a carbon offset provider based on the type of project (e.g. biochar, mineralization) that most aligns with their business ethos.
Today, remote work is the exception rather than the norm. “Flexible work” arrangements tend to be seen as a perk; a 2018 survey found that only around three per cent of American employees worked from home more than half of the time. And yet the technological infrastructure designed for telecommuting hasn’t gone away. It’s what enables employees to answer e-mails on the subway or draft pre-dawn memos in their kitchens. Jack Nilles dreamed of remote work replacing office work, but the plan backfired: using advanced telecommunications technologies, we now work from home while also commuting. We work everywhere.
As spring gives way to summer, and we enter the uncertain second phase of the coronavirus pandemic, it’s unclear when, or whether, knowledge workers will return to their offices.
The machine learning lifecycle begins with data warehousing, ETL pipelining, and model training. At Algorithmia, we focus on the next stages in the lifecycle: deployment, management, and operations. Machine learning deployment plays a critical part in ensuring a model performs well, both now and in the future, but it is also vitally important to understand model monitoring and model drift to that same end.
By monitoring for model drift, you can tell if your model is getting worse over time. For example you can monitor if your model’s accuracy takes a hit after deploying a new model.