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.

May 27, 2022

In today’s Duke AI Health Friday Roundup: avoiding the Turing Trap in AI; monkeypox emerges in US, Europe; roadmap for better western blot data; patent law on collision course with AI; individual variability may still confound mouse models; firearms lead causes of death for children in 2020; EMA puts hold on generics due to dubious bioequivalence studies; retracing the path that let COVID jump from minks to humans; move toward Medicare Advantage plans has implications for availability of data; much more:

Deep Breaths

“But the [von Hemholtz] equation doesn’t work in practice. A pipe’s fundamental tone always sounds lower than the pipe’s length suggests it should according to Helmholtz’s formula. Fixing this problem requires adding an “end correction” to the equation. In the case of open-ended pipes such as flutes and those of organs, the end correction is 0.6 times the radius of the pipe. Why this was, nobody could figure out.” A story in Science News by Bas den Hond explores a minor but persistent mystery behind the tuning of organ pipes and finds some interesting quirks of physics.
It’s a bird, it’s a plane, it’s a… salamander? This article (with accompanying video) by Science’s Erik Stokstad explains why some salamanders are able to launch themselves safely from seemingly dangerous heights.

AI, STATISTICS & DATA SCIENCE

“When AI augments human capabilities, enabling people to do things they never could before, then humans and machines are complements. Complementarity implies that people remain indispensable for value creation and retain bargaining power in labor markets and in political decision-making. In contrast, when AI replicates and automates existing human capabilities, machines become better substitutes for human labor and workers lose economic and political bargaining power.” An essay by Erik Brynjolfsson published in Daedalus makes he case for rejecting the “Turing Trap” when evaluating artificial intelligence, and, rather than conceiving of AI as something that replaces human intelligence, looking for ways that it can be used to augment human intelligence.
“Some technologists, including some at DeepMind, think that one day humans will develop “broader” AI systems that will be able to function as well as or even better than humans. Though some call this artificial general intelligence, others say it is like “belief in magic.“ Many top researchers, such as Meta’s chief AI scientist Yann LeCun, question whether it is even possible at all.” Recent news about advances in AI capabilities courtesy of DeepMind’s “Gato” model have some AI researchers pushing back on some of the more sweeping claims percolating through the media. MIT Technology Review’s Melissa Heikkilä has the story.
“We developed clinically relevant indicators for identifying ACEs [adverse childhood experiences] using the EHRs of mothers and children presenting to general practices and hospital admissions. Over 70% of ACEs were identified via maternal records and were recorded in primary care by GPs within 2 years of birth, reinforcing the importance of reviewing parental and carer records to inform clinical responses to children. A research article published in Lancet Digital Health by Syed and colleagues describes a study in which data from the electronic health records of mothers and children were paired in order to identify potential “adverse childhood experiences.”
“Patent law is based on the assumption that inventors are human; it currently struggles to deal with an inventor that is a machine. Courts around the world are wrestling with this problem now as patent applications naming an AI system as the inventor have been lodged in more than 100 countries…If courts and governments decide that AI-made inventions cannot be patented, the implications could be huge.” A commentary published in Nature by George and Walsh proposes some potential solutions to legal conundrums that emerge when artificial intelligence applications create patentable intellectual property.

BASIC SCIENCE, CLINICAL RESEARCH & PUBLIC HEALTH

“Despite the fact that they’ve been raised in the same conditions and their genetics are nearly identical, each mouse will invariably have somewhat different experiences. Mice are also not automatons and can be expected to vary their behavior from time to time. All of that may set limits on how well we can expect behavioral experiments to replicate.” Ars Technica’s John Timmer reports on a recent study by von Kortzfleisch and colleagues published in PLOS Biology that suggests that at least some of the issues related to replicability (or the lack thereof) in mouse-model studies may stem from individual variation among the inbred strains typically used in these experiments. Meanwhile, in a Nature technology feature, Jyoti Madhusoodanan points out the impact that variations in diet can have on animal experiments.
“When it comes to epidemics, people tend to fight the last war. During the West African Ebola outbreak of 2014, American experts had to quell waves of undue paranoia, which likely contributed to the initial downplaying of the coronavirus. Now, because the U.S. catastrophically underestimated COVID, many Americans are panicking about monkeypox and reflexively distrusting any reassuring official statements.” At The Atlantic, Ed Yong examines news of recent outbreaks of monkeypox in Europe and the US, and places it in the context of an ongoing (but different) pandemic.
“…many questions remain: When, and in whom, did the variant first emerge? How did a taxidermist with no connection to the farm contract it? Could there be a link between the Michigan mink outbreak and a white-tailed deer variant that scientists recently discovered in neighboring Ontario?” The New York Times’ Emily Anthes explores what exactly transpired when a strain of COVID apparently made the leap from minks to humans at a mink farm in Michigan.
“Taken together, our results indicate that GI disorders are characterized by a complex network of associations between microbes and host genes. Although these associations can be disease-specific, we find cases where the same microbial taxon is associated with different host genes in each disease, and vice-versa: cases where the same host pathway is associated with different microbes in each disease.” In an article published in Nature Microbiology by Priya and colleagues, the authors describe a machine-learning analysis that identified connections between human (“host”) gene regulation and disease-causing gut flora.
“Many with sub-Saharan ancestry have a copy of a variant of the gene APOL1 inherited from each parent, which puts them at high risk. Researchers have known for a decade that APOL1 is one of the most powerful genes underlying a common human disease….But there is hope now that much of this suffering can be alleviated.” An article by the New York Times’ Gina Kolata describes recent efforts to apply genetic testing to help guide and target treatment for kidney disease among patients with sub-Saharan African ancestry – as well as the potential complications of such an approach.

COMMUNICATIONS & DIGITAL SOCIETY

“We present detailed descriptions and visual examples to help scientists, peer reviewers and editors to publish more informative western blot figures and methods. Additional resources include a toolbox to help scientists produce more reproducible western blot data, teaching slides in English and Spanish and an antibody reporting template.” A review article by Kroon and colleagues published in BioRxiv examines the reporting of results from western blot protein assays in the peer-reviewed biomedical literature, and finds that such findings are often lacking in detail and adequate provenance (H/T @RetractionWatch).
“Strictly rigorous and reproducible research practices are herewith a sine qua non, but the appropriate and truthful communication of science, its methods, results and pitfalls, is just as important for enhancing research credibility. It is a scientist’s duty to devote appropriate efforts toward good science communication.” An article recently published in BMC Research Notes by Ciubotariu and Bosch presents a framework for improving science communication.
“Researchers in China have lost a 2019 paper on sedation in people undergoing cardiac surgery after readers complained that the authors had failed to properly register the trial.” Retraction Watch has the story of what appears to be a particularly brazen bit of publication misbehavior – a group of authors who used a different trial’s registry number to publish their own (unregistered) trial results.

POLICY

“In the latest dust-up over the safety of medicines, the European Medicines Agency recommended suspending marketing authorization for dozens of generic drugs after finding problems with tests conducted by a contract research organization.” In an article for STAT News (log-in required), Ed Silverman reports on the European Medicines Agency’s recent suspension of a number of generic drugs whose approval rested on bioequivalence studies whose rigor and trustworthiness have now come into question.
“…the reasons for the increase are unclear, and it cannot be assumed that firearm-related mortality will later revert to prepandemic levels. Regardless, the increasing firearm-related mortality reflects a longer-term trend and shows that we continue to fail to protect our youth from a preventable cause of death.” A letter published in the New England Journal of Medicine by Goldstick and colleagues reports on CDC data showing that as of 2020, firearms have overtaken car accidents as the leading cause of child and adolescent death in the United States.
“Too many organizations develop such significant reserves that the statement, “no margin, no mission,” starts to feel extreme and disconnected. For organizations with the market power to extract it, no margin ever seems to be enough. Prudence is replaced with risk-aversion. There is such a thing as too much of a good thing and all of us should ask, ‘how much is too much?’” An opinion article published in Forbes by Sachin Jain examines the question of how much is enough – or too much – when it comes to protecting “margin” in health system budgets.
“Federal policy makers would be wise to plan now for these looming challenges created by the impending transformation of Medicare into a system primarily consisting of competing for-profit health organizations. In addition to technical fixes, such as creating a more sustainable approach to paying Medicare Advantage plans, federal officials should consider how to compensate for the vital information about the US health care system that could be lost with the decline of traditional Medicare.” A viewpoint article published in JAMA by Jacobson and Blumenthal suggests that as Medicare Advantage plans come to predominate over traditional Medicare plans, the increasing spottiness or opacity of data under the former may pose some new challenges.