Stanford researchers lay out a framework for engineers as they expand and build new medical artificial intelligence models.

A patient lies on the operating table as the surgical team reaches an impasse. They can’t find the intestinal rupture. A surgeon asks aloud: “Check whether we missed a view of any intestinal section in the visual feed of the last 15 minutes.”

An artificial intelligence medical assistant gets to work reviewing the patient’s past scans and highlighting video streams of the procedure in real time. It alerts the team when they’ve skipped a step in the procedure and reads out relevant medical literature when surgeons encounter a rare anatomical phenomenon.

Doctors across all disciplines, with assistance from artificial intelligence, may soon have the ability to quickly consult a patient’s entire medical file against the backdrop of all medical healthcare data and every published piece of medical literature online. This potential versatility in the doctor’s office is only now possible due to the latest generation of AI models.

“We see a paradigm shift coming in the field of medical AI,” said Jure Leskovec, professor of computer science at Stanford Engineering. “Previously, medical AI models could only address very small, narrow pieces of the health care puzzle. Now we are entering a new era, where it’s much more about larger pieces of the puzzle in this high stakes field.”

Stanford researchers and their collaborators describe generalist medical artificial intelligence, or GMAI, as a new class of medical AI models that are knowledgeable, flexible, and reusable across many medical applications and data types. Their perspective on this advance is published in the issue of Nature.

Leskovec and his collaborators chronicle how GMAI will interpret varying combinations of data from imaging, electronic health records, lab results, genomics, and medical text well beyond the abilities of concurrent models like ChatGPT. These GMAI models will provide spoken explanations, offer recommendations, draw sketches, and annotate images.

“A lot of inefficiencies and errors that happen in medicine today occur because of the hyper-specialization of human doctors and the slow and spotty flow of information,” said co-first author Michael Moor, an MD and now postdoctoral scholar at Stanford Engineering. “The potential impact of generalist medical AI models could be profound because they wouldn’t be just an expert in their own narrow area, but would have more abilities across specialties.”

Medicine without borders

Of the more than 500 AI models for clinical medicine approved by the FDA, most only perform one or two narrow tasks, such as scanning a chest X-ray for signs of pneumonia. But recent advances in foundation model research promise to solve more diverse and challenging tasks.

“The exciting and the groundbreaking part is that generalist medical AI models will be able to ingest different types of medical information – for example, imaging studies, lab results, and genomics data – to then perform tasks that we instruct them to do on the fly,” said Leskovec.

“We expect to see a significant change in the way medical AI will operate,” continued Moor. “Next, we will have devices that, rather than doing just a single task, can do maybe a thousand tasks, some of which were not even anticipated during model development.”

The authors, which also include Oishi Banerjee and Pranav Rajpurkar from Harvard University, Harlan Krumholz from Yale, Zahra Shakeri Hossein Abad from University of Toronto, and Eric Topol at the Scripps Research Translational Institute, outline how GMAI could tackle a variety of applications from chatbots with patients, to note-taking, all the way to bedside decision support for doctors.

In the radiology department, the authors propose, models could draft radiology reports that visually point out abnormalities, while taking the patient’s history into account. Radiologists could improve their understanding of cases by chatting with GMAI models: “Can you highlight any new multiple sclerosis lesions that were not present in the previous image?”

In their paper, the scientists describe additional requirements and capabilities that are needed to develop GMAI into a trustworthy technology. They point out that the model needs to consume all of the personal medical data, as well as historical medical knowledge, and refer to it only when interacting with authorized users. It then needs to be able to hold a conversation with a patient, much like a triage nurse, or doctor to collect new evidence and data or suggest various treatment plans.

Concerns for future development

In their research paper, the co-authors address the implications of a model capable of 1,000 medical assignments with the potential to learn even more. “We think the biggest problem for generalist models in medicine is verification. How do we know that the model is correct – and not just making things up?” Leskovec said.

They point to the flaws already being caught in the ChatGPT language model. Likewise, an AI-generated image of the pope wearing a designer puffy coat is funny. “But if there’s a high-stake scenario and the AI system decides about life and death, verification becomes really important,” said Moor.

The authors continue that safeguarding privacy is also a necessity. “This is a huge problem because with models like ChatGPT and GPT-4, the online community has already identified ways to jailbreak the current safeguards in place,” Moor said.

“Deciphering between the data and social biases also poses a grand challenge for GMAI,” Leskovec added. GMAI models need the ability to focus on signals that are causal for a given disease and ignore spurious signals that only tend to correlate with the outcome.

Assuming that model size is only going to get bigger, Moor points to early research that shows larger models tend to exhibit more social biases than smaller models. “It is the responsibility of the owners and developers of such models and vendors, especially if they’re deploying them in hospitals, to really make sure that those biases are identified and addressed early on,” said Moor.

“The current technology is very promising, but there’s still a lot missing,” Leskovec agreed. “The question is: can we identify current missing pieces, like verification of facts, understanding of biases, and explainability/justification of answers so that we give an agenda for the community on how to make progress to fully realize the profound potential of GMAI?”

Source: Stanford University

SpaceX headed by billionaire Elon Musk has finally received an official license to launch the Starship rocket to space for the first time in history. The license will be valid for five years.

The license was granted by the U.S. Federal Aviation Administration last Friday. This license was the final frontier that limited the progress of testing the novel launch technology in practice.

The launch should happen tomorrow, April 17. The precise hour was not disclosed, but it should happen from 5:30 a.m. to 2 p.m. Central time. It will be performed from the SpaceX Starbase facility in Boca Chica, Texas – this schedule has been confirmed by the company itself on Twitter.

Targeting as soon as Monday, April 17 for the first flight test of a fully integrated Starship and Super Heavy rocket from Starbase in Texas → https://t.co/bJFjLCiTbK pic.twitter.com/Ry25Uuvknh

— SpaceX (@SpaceX) April 14, 2023

“After a comprehensive license evaluation process, the FAA determined SpaceX met all safety, environmental, policy, payload, airspace integration and financial responsibility requirements,” stated FAA in its official announcement.

Previously, SpaceX targeted earlier dates for the launch of its Starship rocket, but the licensing process delayed the original plans.  The rocket is expected to demonstrate its orbital flight capabilities.

You can offer your link to a page which is relevant to the topic of this post.

German defense industry company Rheinmetall said it was attacked by hackers, but assured that its military business was not affected.

According to the company’s spokesman, the cyber attack was directed against its division dealing with industrial non-military customers, mainly working in the automotive sector.

“Rheinmetall is currently investigating the extent of the damages and is in close contact with the relevant authorities,” commented representatives of the company.

State prosecutors are currently investigating the origin of the attack, but do not provide any specific details, or if this incident was related to the company’s support for Ukraine.

Rheinmetall AG is a German technology group that specializes in security and mobility solutions. The company has two main business segments: Rheinmetall Automotive and Rheinmetall Defence.

Rheinmetall Automotive produces components and systems for the automotive industry, including engine systems, emission reduction technologies, and electric mobility solutions. Rheinmetall Defence offers a range of products and services for the defense sector, including armored vehicles, weapons systems, and simulation and training systems.

You can offer your link to a page which is relevant to the topic of this post.

Researchers have developed a way to create photonic time crystals, showing that these bizarre, artificial materials amplify the light that shines on them. These findings could lead to more efficient and robust wireless communications and significantly improved lasers.

Time crystals were first conceived by Nobel laureate Frank Wilczek in 2012. While conventional natural crystals have a structural pattern that repeats in space, in a time crystal, the pattern repeats in time instead.

Although some physicists were initially skeptical that time crystals could exist, recent experiments have succeeded in creating them. Last year, researchers at Aalto University’s Low Temperature Laboratory created paired time crystals that could be useful for quantum devices.

Now, another team has made photonic time crystals, which are temporal analogues of conventional optical materials. The researchers created photonic time crystals that operate at microwave frequencies and showed that they can amplify electromagnetic waves.

This ability has potential applications in various technologies, including wireless communication, integrated circuits, and lasers. 

So far, research on photonic time crystals has focused on bulk materials – that is, three-dimensional structures. This has proven enormously challenging, and the experiments haven’t gotten past model systems with no practical applications.

Now, the team including researchers from Aalto University, the Karlsruhe Institute of Technology (KIT), and Stanford University tried a new approach: building a two-dimensional photonic time crystal.

‘We found that reducing the dimensionality from a 3D to a 2D structure made the implementation significantly easier, which made it possible to realise photonic time crystals in reality,’ says Xuchen Wang, the study’s lead author, who was a doctoral student at Aalto and is currently at KIT.

The new approach enabled the team to fabricate a photonic time crystal and experimentally verify the theoretical predictions about its behaviour. ‘We demonstrated for the first time that photonic time crystals can amplify incident light with high gain,’ says Wang.

‘In a photonic time crystal, the photons are arranged in a pattern that repeats over time. This means that the photons in the crystal are synchronized and coherent, which can lead to constructive interference and amplification of the light,’ explains Wang. 

Two-dimensional photonic time crystals have a range of potential applications. By amplifying electromagnetic waves, they could make wireless transmitters and receivers more powerful or more efficient.

Wang points out that coating surfaces with 2D photonic time crystals could also help with signal decay, which is a significant problem in wireless transmission. Photonic time crystals could also simplify laser designs by removing the need for bulk mirrors that are typically used in laser cavities.

Another application emerges from the finding that 2D photonic time crystals don’t just amplify electromagnetic waves that hit them from surrounding space but also waves travelling along surfaces. Surface waves are used for communication between electronic components in integrated circuits.

‘When a surface wave propagates, it suffers from material losses, and the signal strength is reduced. With 2D photonic time crystals integrated into the system, the surface wave can be amplified, and communication efficiency enhanced,’ says Professor Viktar Asadchy, who has generated the idea of 2D photonic time crystals.

Source: University of Amsterdam

You can offer your link to a page which is relevant to the topic of this post.

A new study’s finding that urban light pollution may disrupt the winter dormancy period for mosquitoes that transmit West Nile virus could be considered both good and bad.

The good news is that the disease-carrying pests may not survive the winter if their plans to fatten up are foiled. The bad news is their dormancy period, known as diapause, may simply be delayed – meaning they’re biting humans and animals longer into the fall. 

“We see the highest levels of West Nile virus transmission in Ohio’s late summer and early fall. If you have mosquitoes postponing or delaying diapause and continuing to be active longer in the year, that’s at a time when the mosquitoes are most likely to be infected with West Nile virus and people could be at greatest risk of contracting it,” said Megan Meuti, senior author of the study and an assistant professor of entomology at The Ohio State University. 

This study and earlier findings by Meuti and her colleagues are among the first to show that artificial light at night could significantly impact mosquito behavior – including effects that aren’t necessarily predictable. 

“We’re finding that the same urban light at night can have very different effects under different seasonal contexts,” she said. 

Meuti conducted the study with first author Matthew Wolkoff and Lydia Fyie, both PhD candidates in entomology at Ohio State. The research was published recently in the journal Insects. 

Diapause for female Northern house mosquitoes (Culex pipiens) is not quite a winter slumber, but rather a period of dormancy when the insects live in caves, culverts, sheds and other semi-protected locations. Prior to winter’s arrival, mosquitoes convert sugary sources, such as plant nectar, into fat.

As days get longer, females begin foraging for blood meals to enable egg production. Some get infected with West Nile virus by feeding on infected birds, and later transmit the virus when they feed on people, horses and other mammals. 

This study builds upon two previous findings from Meuti’s lab: For her dissertation, Meuti found that circadian clock genes differ between diapausing and non-diapausing mosquitoes, strongly suggesting that day length dictates when diapause should start.

And more recent work led by Fyie found that female mosquitoes exposed to dim light at night averted diapause and became reproductively active – even when short days indicated they should be dormant. 

In the current study authored by Wolkoff, the researchers pursued both lines of inquiry, comparing daily activity and nutrient accumulation by mosquitoes reared in two lab conditions – long days mimicking the insects’ active season and short days that induced dormancy – with and without exposure to artificial light at night. 

The study provided more evidence associated with a circadian pattern to mosquito behavior, showing that insects’ activity decreases during diapause, but the circadian rhythmicity of that activity is sustained even during this dormant period. 

The introduction of artificial light at night was found to affect those activity patterns and to influence mosquitoes’ acquisition of nutrient reserves needed for fattening up and weathering winter temperatures. 

Exposure to light pollution suppressed the amount of water-soluble carbohydrates – sugars that are an essential food source during winter – that were accumulated by mosquitoes in both long- and short-day conditions.

Patterns of accumulation of the sugar glycogen were reversed by exposure to artificial light at night: Under normal conditions, non-dormant mosquitoes had lots of glycogen in their bodies but diapausing bugs did not – but in mosquitoes subjected to light pollution, the long-day mosquitoes didn’t accumulate much glycogen and short-day mosquitoes showed an increase in glycogen accumulation.

The researchers observed consistent trends in activity-related effects of light at night, with slight increased activity among the dormant mosquitoes and slightly suppressed activity among long-day mosquitoes expected to be busy looking for food.

Though the findings weren’t statistically significant, Wolkoff said the combined observations suggest light pollution causes mosquitoes to ward off diapause – perhaps by scrambling signals from their circadian clock.

“This could be bad for mammals in the short term because mosquitoes are potentially biting us later in the season, but it could also be bad for mosquitoes in the long term because they might be failing to fully engage in preparatory activities they need to survive the winter during diapause, and that might reduce their survival rate,” Wolkoff said. 

The researchers plan to carry out field studies to see if these lab findings hold true in the wild.

Source: Ohio State University

After nearly a century of scientific inquiry, scientists have finally characterized a key component in the substance called melanin responsible for giving countless living organisms their color. 

In the study, published online in the journal Nature Chemistry, an international team of researchers isolated a key molecule involved in the synthesis of melanin, a substance in the human body that produces pigmentation in the hair and skin and protects the cells from being damaged by ultraviolet radiation from the sun.

The molecule they studied has many physical properties of eumelanin, a type of melanin that typically produces only black and brown pigments. 

Despite what researchers know about melanin, its chemical structure has remained elusive, said Bern Kohler, an Ohio Eminent Scholar and professor of chemistry and biochemistry at The Ohio State University, one of three senior authors.  

“Melanin is as plain as the nose on our face and we still don’t know exactly what it’s made of and how it works,” said Kohler. “It’s thought to be a material made of large numbers of interacting components, and so what my collaborators and I are trying to get at is, what are melanin’s underlying chemical units and what are the interactions that give rise to its properties?”

In the lab, melanin can be made naturally from just a few simple chemical substances that combine or react to form the pigmented material. Thanks to the efforts of Jean-Philip Lumb of McGill University and Lluis Blancafort of the University of Girona, the team was able to successfully synthesize and analyze one of its molecule-sized ingredients. They found that it had many of the same properties as the final product melanin, even without further transformation.

“Melanin is like a complex dish that you cook from just a few ingredients,” said Kohler. “Given that melanin granules in skin contain billions of atoms, it was surprising to see melanin-like properties emerge in a molecule containing only a few dozen atoms.”

As a spectroscopist – or a scientist who investigates the interactions between matter and light – Kohler said what he finds most remarkable is the brilliant green color of the melanin-like molecule, which is able to absorb deep red light. 

“That’s very hard for a small molecule to do,” said Kohler. “This is one of the smallest organic molecules known that can absorb long wavelengths all the way out into the infrared.”

Its unusual optical and magnetic properties make the melanin-mimicking molecule an attractive prospect for advancing bioelectronics research, a field that aims to connect electronics and living biological materials to fabricate new technologies or therapies for medical treatments, said Kohler.

The study results also suggest that harnessing the power of melanin could be used to change the way humans harvest energy from our environment.

“Like solar energy conversion, part of our strategy for alternative energy is to absorb photons of light from across the solar spectrum,” Kohler said. “That’s what eumelanin does, rather naturally.”

Future research will aim to study melanin-like properties in similar kinds of molecules. Such strides could offer greater accessibility to synthetic biomaterial alternatives, but in the short term, Kohler imagines the team’s work uncovering a melanin-like molecule will inspire other labs to look more closely at how melanin’s properties may emerge from other unexpectedly small molecular units. 

“Our team is excited that further progress at understanding the structure of melanin can succeed by studying tailorable small molecules like the one we write about in this paper,” he said. “The chemistry is still tremendously complex, but interest should grow in small molecules as melanin models and the promise they hold for generating new, melanin-inspired materials.”

Source: Ohio State University

You can offer your link to a page which is relevant to the topic of this post.