AI Health

Friday Roundup

The AI Health Friday Roundup highlights the week’s news and publications related to artificial intelligence, data science, public health, and clinical research.

December 17, 2021

In today’s Roundup: looking ahead to the next pandemic; best data visualizations of 2021; health AI for the Global South; meta-analysis sharpens focus on ‘long COVID’; diet, gut microflora, and immunotherapy; sorting through Web3 hype; despite progress, chatbots still go off the rails; flattery for dictator still enshrined in scientific literature; the state of scientific peer review in 2021; much more:

Deep Breaths

Restored nineteenth century painting of two Australian nocturnal ground parakeets. Painted by Elizabeth Gould. Image courtesy raxpixel/Wikipedia (CC BY-SA 4.0).
Image credit: original painting: Elizabeth Gould; digital enhancement by raxpixel/Wikipedia (CC BY-SA 4.0).
  • No stranger to our Deep Breaths section, Yo-Yo Ma rings in 2022 with a cello rendition of “Auld Lang Syne.”
  • “It’s hard to imagine a more elusive bird to track than the night parrot. The nocturnal, ground-dwelling birds shelter amid thick clumps of dry, spiky grass in Australia’s most isolated and harshest regions — some more than 1,000 miles from the closest city…Until Mr. Young’s discovery, almost everything scientists knew about the night parrot came from amateur ornithologists’ 19th-century diary entries and a small number of museum specimens.” The New York Times’ Anthony Ham brings us the strange saga of the rediscovery of an elusive Australian parrot, once thought extinct.
  • Over the recent holiday break, we bid farewell to a pair of major figures in 20th-century science: evolutionary biologist O. Wilson and paleoanthropologist Richard Leakey.

AI, STATISTICS & DATA SCIENCE

Photograph of river rapids with snow-covered, cloud-shrouded mountain peak in the immediate background. Image credit: Diego Jimenez/Unsplash https://unsplash.com/photos/XHDtPRj535A
Image credit: Diego Jimenez/Unsplash
  • We’re starting off the year with some great dataviz roundups from the previous year. At Flowing Data, Nathan Yau presents his personal best-of-2021 projects in data visualization, while at her blog, scientist Helena Jambor runs down notable trends in scientific dataviz from 2021. (Interestingly, there is no overlap between Yau and Jambor’s examples, although the latter is more narrowly focused on biological sciences. Yau’s best-of list also contains one we’d stumbled across ourselves: Sam Learner’s fascinating River Runner project that allows you to trace the fate of an individual raindrop from its initial fall to its ultimate destination in lake or ocean.
  • An article published in Lancet Digital Health by Mohammad and colleagues describes a study that evaluated the accuracy of neural network in predicting 1-year mortality and admission for heart failure following myocardial infarction in a patient cohort drawn from the SWEDEHEART registry.
  • If you missed the University of Michigan’s 2021 AI Symposium, no worries – all sessions are available as a YouTube playlist.
  • “Over the past 60 years, artificial intelligence (AI) has made significant progress, but most of its benefits have failed to make a significant impact within the Global South. Current practices that have led to biased systems will prevent AI from being actualized unless significant efforts are made to change them.” A perspective article by Chinasa T. Okolo at Cell Patterns delves into the issues emerging from attempts to deploy healthcare AI applications across the Global South.
  • “Similar to previous NDAAs, the 2022 version contains a number of provisions directly relevant to artificial intelligence (AI). We identified and described the provisions related to AI and other activities or technologies that could impact U.S. advancement of AI in this 2,165-page bill.” Stanford University’s Human Centered Artificial Intelligence institute has combed through the most recent National Defense Authorization Act to highlight all of the provisions related to artificial intelligence.
  • Wishart says that the algorithm could help law enforcement and forensic chemists identify novel psychoactive substances — a process that could otherwise take up to several months….But critics argue the proactive criminalization of drugs would do little to address the main drivers behind overdose deaths. Moreover, criminalization has historically been socially damaging…” At Undark, Doug Johnson reports on efforts to use AI algorithms to get the jump on designers of illegal drugs, who often attempt to skirt criminal statutes by making minor alterations to the chemical formulas of synthetic drugs.
  • “Using statistical patterns to create the illusion of human-like conversation is fundamentally different from understanding what is being said…This is far more than a philosophical debate about the meaning of the word understanding….Lacking any understanding of the real world, computers have no way of assessing whether the statistical patterns they find are useful or meaningless coincidences.” An article by Gary Smith at Mind Matters News considers recent progress in natural-language chatbots, judges them, and finds them (still) wanting (H/T @yudapearl).

BASIC SCIENCE, CLINICAL RESEARCH & PUBLIC HEALTH

Close-up photograph of shrimplike Northern Krill. Image credit: Oystein Paulsen via Wikipedia (CC BY-SA 3.0)
Image credit: Oystein Paulsen via Wikipedia (CC BY-SA 3.0)
  • “…blood triglyceride levels were reduced by 26.0% in the ω-3–PL/FFA group and 15.1% in the placebo group at 12 weeks, for a significant mean treatment difference that persisted at 26 weeks. ω-3–PL/FFA was well tolerated, with a safety profile similar to that of placebo.” A paper published by Mozaffarian and colleagues in JAMA Network Open reports pooled results from a pair of randomized, placebo controlled trials that evaluated the effectiveness of a new omega-3 oil derived from krill in lowering cholesterol levels in patients with severely elevated levels of triglycerides.
  • “A significant proportion of individuals experience persistent fatigue and/or cognitive impairment following resolution of acute COVID-19. The frequency and debilitating nature of the foregoing symptoms provides the impetus to characterize the underlying neurobiological substrates and how to best treat these phenomena.” A timely systematic review and meta-analysis published in Brain, Behavior, and Immunity by Ceban and colleagues sharpens the focus on symptoms, especially fatigue and cognitive impairment, associated with the complex phenomenon known as “post-COVID syndrome.” (H/T @handvanish)
  • “Our data shed light on how lineages and fates are established during cerebral organoid formation. More broadly, our techniques can be adapted in any iPSC-derived culture system to dissect lineage alterations during normal or perturbed development.” A paper recently published in Nature Methods by He and colleagues traces a new approach for tracing the genetic lineage of cerebral organoids (artificially created blobs of tissue that mimic the functioning of the larger organs from which they are derived).
  • “In an observational study, the researchers found that melanoma patients reporting high fiber (prebiotic) consumption had a better response to checkpoint inhibitor immunotherapy compared with those patients reporting a low-fiber diet. The most marked benefit was observed for those patients reporting a combination of high fiber consumption and no use of over-the-counter probiotic supplements.” A research article published recently in Science by Spencer and colleagues examines interactions between diet and gut microflora – the bacteria that naturally inhabit the human digestive tract – and how those interactions can effect the working of immunotherapy for melanoma.
  • “The reduction in the community prevalence of yaws was greater with three rounds of mass administration of azithromycin at 6-month intervals than with one round of mass administration of azithromycin followed by two rounds of targeted treatment.” A paper published by John and colleagues in the New England Journal of Medicine describes results from a cluster-randomized trial of different azithromycin dosing regimens for the bacterial disease yaws.
  • “It has been widely noted that some people are more attractive to mosquitoes than others, but the mechanistic basis of this phenomenon is poorly understood. Here we tested mosquito attraction to skin odor collected from human subjects and identified people who are exceptionally attractive or unattractive to mosquitoes.” A preprint article by De Obaldia and colleagues, available at bioRxiv, may shed some light on why mosquitoes sometimes seem to prefer some humans over others for snacking purposes (@DerekLowe).
  • “Despite its small size, the results in the study are remarkably consistent. Not a single rapid antigen test detected the virus until nearly two days after the initial positive PCR result. Additionally, the cases of infection from people who had received false negative results could raise alarm bells.” At STAT News, Matthew Herper unpacks findings from a recently posted preprint (not yet peer-reviewed) that may indicate some rapid antigen tests aren’t detecting the presence of COVID with the first days of an infection.
  • “So there’s a shortage of the stuff, that’s used to make the stuff, that’s used to make the stuff, that’s used to make two of the starting materials for Paxlovid. And that’s just one of the reagents….That’s how it goes in the fine chemical business – there’s a compound that no one really cares much about, until they do and they care hugely, and then suddenly no one cares about it again, until some other bizarre reason emerges to put it back into demand.” At In the Pipeline, Derek Lowe explains why the current shortages of the antiviral COVID therapy Paxlovid are proving difficult to resolve quickly.

COMMUNICATIONS & DIGITAL SOCIETY

Man standing on a beach with back turned, silhouetted against a colorful sunset sky. Image credit: Maciej Pienczewski/Unsplash
Image credit: Maciej Pienczewski/Unsplash
  • “Twilight has resulted in two distinct, simultaneous, and unique roles in my life. The first is that of a neurologist with decades of experience with DLB and a large practice of patients with the disorder. The second is that of a patient with DLB. They are two very different experiences. An analogy would be to look at a hurricane from far away versus looking at it from within. The inner and outer visions result in very different perspectives.” In an amazing personal essay published at Behavioral Neurology and Psychiatry (made available through Continuum), neurologist Daniel Alejandro Druback describes his self-diagnosis of Lewy Body Dementia (LBD), a condition that he renamed “Twilight.”
  • “If you’ve been on Twitter the last few days, you’ve seen how heated the debate has gotten over all things decentralized. Even the terms are complicated: Web3, crypto, blockchain and decentralization all get lumped together despite often meaning hugely different things. And the debate seems to often veer into ideology rather than technology.” At Protocol, David Pierce tries to sort out the signal from the noise in the hypestorm surrounding the advent of the so-called Web3.
  • “Elena Ceaușescu was celebrated by state propaganda under her husband’s regime as a world-famous chemistry researcher, despite having no credible qualifications. The researchers say some of her work is still being cited and accessed, even though she was barely literate in science and unable to recognise basic formulas taught to first-year chemistry students.” The Guardian’s Melissa Davey delves into a long-delayed reckoning with an international legacy of unearned and undeserved academic prestige and flattery that was showered upon Elena Ceauşescu, the wife of the Romanian dictator, whose ostensible scientific credentials were wholly bogus (H/T @RetractionWatch).
  • “Peer review at journals is a critical part of science, but we’re spectacularly unscientific about it. Yet progress on addressing all the unanswered questions about it is still painfully slow….Take biomedicine, where most of the experimental research on peer review has been done. A systematic review published in 2021 only found 3* more trials than a systematic review with a search for studies 5 years earlier. Across the whole of science, we can’t even count on each year bringing us a solid trial that advances knowledge enough to improve journal practices.” At her PLOS blog Absolutely Maybe, Hilda Bastian reflects on the state of peer review in 2021.

POLICY

Man facing camera while looking through giant binoculars mounted on a stand or tripod. Image credit: Mostafa Meraji/Unsplash
Image credit: Mostafa Meraji/Unsplash
  • “After previous infectious disease threats, the US quickly forgot and failed to institute necessary reforms. That pattern must change with the COVID-19 pandemic. Without a strategic plan for the “new normal” with endemic COVID-19, more people in the US will unnecessarily experience morbidity and mortality, health inequities will widen, and trillions will be lost from the US economy. This time, the nation must learn and prepare effectively for the future.” A Viewpoint article by Emanuel and colleagues, just published at JAMA, makes an urgent case for rapidly assimilating the lessons of the COVID to prepare for coming pandemic threats.
  • “NYULH took an uncommon, value-driven approach in ensuring true integration, including a unified clinical and administrative governance structure supported by a robust electronic information system, including a common EHR and cost-accounting system. This focus on robust integration was balanced with identification of local opportunities, implementation of site-specific quality improvement interventions, and a systemwide adoption of some of these novel approaches.” A health policy analysis published by Wang, Arnold, and Jones in JAMA Network Open examines the effects on quality of care and patient safety following a hospital merger (H/T @leorahorwitzmd).
  • “The eight risk communication mistakes in this commentary aren’t necessarily the biggest challenges public health officials and experts face—maybe not even their biggest risk communication challenges. But they are likely to be among the most remediable of their risk communication challenges, since they stem from their own behavior. I think these mistakes keep happening, they do real damage, but they can be remedied—so revisiting them isn’t just backward-looking.” In a post for the University of Minnesota’s Center for Infectious Disease Research and Policy, Peter M. Sandman dissects public-health communication fumbles during the COVID pandemic.
  • “After previous infectious disease threats, the US quickly forgot and failed to institute necessary reforms. That pattern must change with the COVID-19 pandemic. Without a strategic plan for the “new normal” with endemic COVID-19, more people in the US will unnecessarily experience morbidity and mortality, health inequities will widen, and trillions will be lost from the US economy. This time, the nation must learn and prepare effectively for the future.” A Viewpoint article by Emanuel and colleagues, just published at JAMA, makes an urgent case for rapidly assimilating the lessons of the COVID to prepare for coming pandemic threats.