In today’s Roundup: the math of moving people through the Louvre; CAR-T for cardiac fibrosis; more probes for the proteome; cybercrooks target IT for smaller clinics; UK debuts “data hubs”; materials science grapples with machine learning bias; the health toll of hearing loss; DNA testing service accused of fudging results; some cancer treatments may be “working” through off-target effects; the heartbreak of dichotomania; overhauling medical education; much more:
AI, Statistics, & Data Science
- At Duke’s Research Blog, Robin Smith describes the work of a prize-winning team of Duke undergrads who used computer simulations to develop an optimal method for evacuating tens of thousands of tourists out of the Louvre quickly and safely (H/T @DukeResearch).
- “My point is that discrete thinking is not simply something people do because they have to make decisions, nor is it something that people do just because they have some incentive to take a stance of certainty. So when we’re talking about the problems of deterministic thinking, or premature collapse of the “wave function” of inferential uncertainty, we really are talking about a failure to incorporate enough of a continuous view of the world in our mental model.” At his blog, statistician Andrew Gelman tackles the scourge of “dichotomania” – the impulse to divide data into discrete groups even when doing so may distort our understanding of underlying phenomena. However, Gelman – who prefers “deterministic thinking” to the somewhat pejorative “dichotomania” – argues that the problem goes beyond statistical practice and is rooted in the fundamental ways humans think about things.
- “No one would argue that the consequences of biased chemical data are as serious as those of biases in facial-recognition software, but they share a similar origin. Researchers should be alert to the potential for bias in their chemical data sets, before it gets baked into a machine.” An editorial in Nature highlights recent research by Jia and colleagues showing that the influence of human biases on methods and data sets can have profound implications for the success of machine learning applications in multiple disciplines—including, in this case, chemistry and materials science.
- We missed this one last month, but this TechCrunch interview with Columbia’s Desmond Patton on why the field of AI needs more social workers is definitely worth a read.
- A research article by Polubriaginof and colleagues published in the Journal of the American Informatics Association examined observational databases and EHR data to assess the quality of information about race and ethnicity. The authors found large proportions of missing data and substantial discrepancies between what was recorded in these repositories versus what patients themselves were able to directly record their own race and ethnicity information (H/T @rusincovitch).
- “Many in the digital medicine community recognize the importance of adopting standards for the format, exchange and security of digitally captured clinical data. Without consistent use of agreed upon standards, digital medicine risks further burdening the healthcare system with untrustworthy and siloed data — but where do we begin?” In a post for Medium, Digital Medicine Society Research Lead Christine Manta takes a dive into the alphabet soup of different standards for capturing, storing, and exchanging digital health data.
- A press release from Health Data Research UK highlights the creation of 7 new National Health Service-funded “data hubs” designed to expedite access to clinical data and accelerate clinical research (H/T @MartinLandray).
Basic Science, Clinical Research & Public Health
- Bloomeberg’s Kristen Brown reports allegations from former employees of DNA testing startup Orig3n that the company, which offered to match genotypes to optimize selection of food, exercise, and more, may have been putting its thumb on the scales to produce desired results after its tests were revealed to be subject to errors (H/T@CecileJannsen).
- Multiple journalists, including Carl Zimmer at the New York Times and Sharon Begley at STAT News, are reporting a recent study by Lin and colleagues published in Science Translational Medicine that suggests a possible reason that so many cancer therapy candidates appear so promising in lab and animal studies, only to fail or disappoint in humans: it may be the off-target toxicities of these compounds that are doing the “heavy lifting” of killing cancer cells, rather than actual targeted pathways. If the results of this study are confirmed, they will have substantial implications for how candidate drugs are evaluated before advancing to clinical studies.
- “In the public imagination, suicide is often understood as the end of a torturous decline caused by depression or another mental illness. But clinicians and researchers know that suicidal crises frequently come on rapidly, escalating from impulse to action within a day, hours, or just minutes.” At Undark, Temma Ehrenfeld examines the push to reconsider how doctors think about and diagnose suicidality.
- “Despite decades of hunting for compounds that researchers can use to modulate the activity of particular proteins in cells, such chemical probes remain few and far between. By one estimate, small-molecule probes are currently available for only 4% of the human proteome. And although many factors drive research priorities, the lack of tools to selectively turn proteins on or off contributes to the inability to study massive tracts of biological space.” A Nature Reviews Drug Discovery News article by Asher Mullard details the potential, as well as the significant efforts, that await Target 2035, an ambitious global effort to develop a comprehensive suite of molecular probes to encompass the entire human proteome.
- A cross-sectional study by Steelesmith and colleagues published last week in JAMA Network Open examines geographic and social influences in U.S. suicide rates at the county level.
- A research letter by Aghajanian and colleagues published in Nature describes a study that successfully used CAR-T immunotherapy to treat cardiac fibrosis in mice, potentially paving the way for new applications of what had previously been better known as a cancer therapy.
- “As hearing declines, loneliness can intensify — and set off a cascade of detrimental health effects. Now considered as hazardous as smoking 15 cigarettes a day, loneliness vastly raises the risks of depression, dementia and early death. Yet the vast majority of people who suffer from hearing loss don't know they have a problem — or don't want to know.” NPR’s Rochelle Sharpe examines the problem of untreated hearing loss among older persons, and finds that it can be much more than a nuisance: untreated hearing loss can bring with it ever worsening loneliness and social isolation – and those in turn entrain a host of other serious health problems.
- “What makes the company’s colors so revolutionary is that they radiate in sunlight, while ordinary “neon” pigments glow only with black lights in the dark. This atomic innovation is what drew artists and industrial designers to the medium. Day-Glo paints are intrinsically technical, a truly Space-Age material.” The LA Times’ Sonja Sharp gives us a guided tour of the history and science of Day-Glo colors – and the quest to reverse-engineer “Saturn Yellow” before certain works of art fade beyond recovery.
Communications & Digital Society
- A post at Axios from the Kaiser Family Foundation’s Drew Altman tries to winnow the hope from the hype when it comes to health technology: “70% of the people we surveyed say they’ve used the internet to research symptoms or learn more about health conditions. And 51% use apps or other tech tools to track their sleep, fitness or diet. But as people’s needs shift from personal information-gathering into the formal health care system, their tech usage begins to fall.”
- From the New York Times Learning Network come this article on Japanese abacus tournaments and the relative merits of learning to do complex arithmetical calculations without a computer or (electronic) calculator.
- “We uncovered a network of at least 16 Facebook pages hiding their white nationalist/white supremacist and racist views behind misleading names, calling users to support policemen, firefighters, or armed forces, and sometimes even victims of terrorism. On Facebook alone, these pages collectively reach an audience of 375,000 people.” The EU Disinfo Lab has released a report this week detailing the layered, complex, and wide-ranging network of social-media activity designed to deceptively boost (and solicit funding for) racist ideologies across Europe (H/T @gchaslot).
- “Beyond the individual victims, the MSPs’ shortcomings have a larger consequence. They foster the spread of ransomware, one of the world’s most common cybercrimes. By failing to provide clients with reliable backups or to maintain their own cybersecurity, and in some cases paying ransoms when alternatives are available, they may in effect reward criminals and give them an incentive to strike again.” ProPublica’s Renee Dudley has a hair-raising account of a new target that’s attracting the attention of cybercriminals – managed service providers, or MSPs, the IT middlemen who provide services for customers that include civic governments and small-to-medium healthcare practices, but who often lack the security measures and resources to fend off determined hacking attempts and ransomware exploits.
- “To be sure, a full Whitman’s Sampler of unpleasant biological contingencies can arrive with age, and death ultimately comes for us all. But the difference between those hard truths and the dominant narrative of old age that we’ve inherited is big enough and persistent enough to account for the expectations gap—and then some.” Really feeling this one: at MIT Technology Review, Joseph F. Coughlin argues that the idea of “old age” as a discrete state that people inevitably enter – a state implicitly marked by decrepitude and incapacity – is artificial, pernicious, and most of all, untrue.
- Viral sloganeering! Keyword squatting! Online misinformation has become such a large and ramifying problem that we’re having to constantly coin new terminology to describe various novel informational pathogens. Poynter has some of the latest.
- Research Triangle International (RTI) has published a meeting summary from a symposium held earlier this year to address ways to build and maintain public trust in science (H/T@BrianSouthwell).
- “…in the last few years, the remarkable progress made by vaccines has gone into reverse, particularly in the West. In Europe, there were about 5,000 cases of measles in 2016, 24,000 in 2017, and 84,000 in 2018. According to recent WHO data, there have already been 90,000 measles cases in the first six months of this year. And the US is experiencing its biggest measles outbreak for 27 years.” A Quartz article by Jonathan Kennedy outlines the public-health perils of backsliding on immunization, as decades of progress against epidemic disease are being eroded by the efforts of anti-vaccination movements.
- An editorial in JAMA by Green, Wayne, and Neilson makes the case for overhauling and modernizing the overall approach to medical education for physicians.
- Writing in the New York Times opinion section, cardiologist Sandeep Jauhar advises physicians who gripe about overregulation to consider the virtues of preemptively cleaning house.
- “Uber and Lyft decimated the yellow taxi industry, sending medallion prices plummeting, and many drivers into deep debt. Some drivers were so despairing they took their own lives. Meanwhile, traffic congestion soared in cities, much of it attributable to the popularity of ride-hailing. The backlash against tech companies like Uber and Lyft seems in line with similar reckonings for Apple, Facebook, Amazon, and Google.” At the Verge, Andrew Hawkins unpacks the implications of California’s Assembly Bill 5, which directly affects the ability of ride-sharing companies like Uber and Lyft to treat workers as contractors rather than employees, and places the heightened regulatory interest in these companies in the context of a larger cultural backlash to Silicon Valley.
- “In the absence of hospital policies, doctors can find it difficult to deal with patients who act inappropriately toward them. They try to empathize with patients who may feel angry, frustrated, or powerless when interacting with the sometimes inefficient and bureaucratic health care system. They know patients are often stressed and frightened, and most physicians feel they have a duty to provide care, even when patients are difficult. So when doctors feel harassed, they often just laugh it off in an attempt to diffuse the situation and avoid offending the patient.” STAT News’ Jacquelyn Corley examines the problem of harassment of hospital staff by patients – and what some health systems are doing about it.