In today’s Duke AI Health Friday Roundup: health and the genetics of circadian rhythms; trust and human-robot interactions; how bias gets built into GANS; Stone Age surgery; probing the limits of scientific education and civic engagement; judge rules against PrEP coverage; prosthetics for memory; arguing for and against including AI in medical training; much more:
AI, STATISTICS & DATA SCIENCE
- “…if language grounding is ‘where robotics and [natural language processing] NLP meet’, then we are interested in ‘where [human-robot interaction] meets NLP’. We propose that language grounding comes with its own variant of trust, which we call trust in language grounding; does the human trust the robot to understand language deeply, in the same sense that humans do, by understanding how all the objects, actions, and abstractions related to each other and how they might look or feel like in the real world?” In a preprint survey paper available from arXiv, Bossens and Evers explore the boundaries of trust between human and machine collaborators.
- “Although artificial intelligence (AI) has immense potential to shape the future of medicine, its place in undergraduate medical education currently is unclear. Numerous arguments exist both for and against including AI in the medical school curriculum. AI likely will affect all medical specialties, perhaps radiology more so than any other.” At Radiology: Artificial Intelligence, Ngo, Nguyen, and van Sonnenberg argue both sides of the case for formally including artificial intelligence as part of medical training for physicians.
- “This led Anandkumar to challenge AI’s reliance on matrix methods. She deduced that to keep an algorithm observant enough to learn amid such chaos, researchers must design it to grasp the algebra of higher dimensions. So she turned to what had long been an underutilized tool in algebra called the tensor. Tensors are like matrices, but they can extend to any dimension, going beyond a matrix’s two dimensions of rows and columns.” An interview at Quanta with computer scientist Anima Anandkumar highlights her efforts to push the field of machine learning beyond its reliance on a ubiquitous mathematical tool.
- “…we observe that truncation, a technique used to generate higher quality images during inference, exacerbates racial imbalances in the data. Lastly, when examining the relationship between image quality and race, we find that the highest perceived visual quality images of a given race come from a distribution where that race is well-represented, and that annotators consistently prefer generated images of white people over those of Black people.” A preprint article available at arXiv by Maluleke and colleagues explores the workings of racial bias in generative adversarial networks (GANS), a type of machine learning widely used for training image-recognition and image-generation applications.
BASIC SCIENCE, CLINICAL RESEARCH & PUBLIC HEALTH
- “Circadian rhythms and sleep are fundamental biological processes integral to human health. Their disruption is associated with detrimental physiological consequences, including cognitive, metabolic, cardiovascular and immunological dysfunctions. Yet many of the molecular underpinnings of sleep regulation in health and disease have remained elusive.” A new article at Nature Reviews Genetics by Lane and colleagues explores the current state of knowledge about the connections between genetics, sleep cycles, and health in humans.
- “A unique form of brain stimulation appears to boost people’s ability to remember new information—by mimicking the way our brains create memories… The ‘memory prosthesis,’ which involves inserting an electrode deep into the brain, also seems to work in people with memory disorders—and is even more effective in people who had poor memory to begin with, according to new research.” An article by MIT Technology Review’s Jessica Hamzelou details the emerging (and still experimental) technology of “brain prostheses,” which, if it pans out, could potentially help patients with memory problems or brain damage (H/T @NitaFarahany).
- “These models, called ‘phantoms’, are made of materials that mimic human bones, soft tissue and organs, and they’ll be fitted with sensors that will measure radiation exposure as they travel to the moon and back. Paul Segars and Ehsan Samei, both researchers at the Carl E. Ravin Advanced Imaging Laboratories at the Duke University School of Medicine, helped develop these phantoms using methods originally created to study how different medical procedures, tools and techniques precisely affect organs throughout the human body.” A news article by Duke’s Michaela Kane points out that while the first of NASA’s upcoming missions to the moon will lack astronauts, the Orion capsule will in fact be occupied by data-collecting “phantoms” that Duke researchers contributed to.
- “Here, however, we report the discovery of skeletal remains of a young individual from Borneo who had the distal third of their left lower leg surgically amputated, probably as a child, at least 31,000 years ago. The individual survived the procedure and lived for another 6–9 years, before their remains were intentionally buried…This unexpectedly early evidence of a successful limb amputation suggests that at least some modern human foraging groups in tropical Asia had developed sophisticated medical knowledge and skills long before the Neolithic farming transition.” In an article published in Nature, Maloney and colleagues describe the archaeological evidence that led a team of researchers to identify what may be the earliest known example of a surgical limb amputation.
COMMUNICATION, Health Equity & Policy
- “In situations like this, one-way communication from scientific authorities may not be sufficient to restore trust—even when it seeks to demonstrate why a particular scientific claim is not a product of illegitimate or discriminatory practices. The obstacle in such cases isn’t an irrational reluctance to consider new information. It’s that the information itself is deemed unreliable, in part because of a poor moral assessment of who or what is conveying it. Addressing these perceived moral failures of scientific or medical institutions may require much more than the mere conveyance of technical information.” A thought-provoking article by Catarina Dutilh Novaes and Silvia Ivani at the Boston Review argues that current approaches to closing gaps between practitioners of science, medicine, and technology development and the general public are hampered by an outdated and inaccurate model of science communication.
- “We have not done nearly enough to prevent or prepare for such potential pandemics. While there are certainly gaps in our scientific defenses, the bigger problem is that leaders at all levels have not been giving these threats anything close to the priority they demand. Ebola and other outbreaks revealed gaping holes in preparedness, serious weaknesses in response, and a range of failures of global and local leadership. This is the neglected dimension of global security.” An early-release version of an upcoming report from the National Academy of Medicine’s Commission on a Global Health Risk Framework for the Future makes an urgent case for better global pandemic preparedness (H/T Gavin Yamey).
- “The settlement, which includes numerous restrictions on how Juul can market its products, resolves one of the biggest legal threats facing the beleaguered company, which still faces nine separate lawsuits from other states. Additionally, Juul faces hundreds of personal lawsuits brought on behalf of teenagers and others who say they became addicted to the company’s vaping products.” At STAT News, the AP’s Matthew Perrone and Dave Collins report on a tentative settlement reached between vaping manufacturer Juul Labs and a group of state attorneys general over the company’s marketing practices.
- STAT News’ Bob Herman reports on a recent ruling by a federal judge that holds that the Affordable Care Act’s mandate for insurers to provide certain kinds of drugs and preventive services are unconstitutional, including coverage requirements for the HIV prophylactic drug combination known as PrEP.