Now, EMBL scientists have combined artificial intelligence (AI) algorithms with two cutting-edge microscopy techniques — an advance that shortens the time for image processing from days to mere seconds, while ensuring that the resulting images are crisp and accurate. The findings are published in Nature Methods.

“Ultimately, we were able to take ‘the best of both worlds’ in this approach,” says Nils Wagner, one of the paper’s two lead authors and now a PhD student at the Technical University of Munich. “AI enabled us to combine different microscopy techniques, so that we could image as fast as light-field microscopy allows and get close to the image resolution of light-sheet microscopy.”

Although light-sheet microscopy and light-field microscopy sound similar, these techniques have different advantages and challenges. Light-field microscopy captures large 3D images that allow researchers to track and measure remarkably fine movements, such as a fish larva’s beating heart, at very high speeds. But this technique produces massive amounts of data, which can take days to process, and the final images usually lack resolution.

Light-sheet microscopy homes in on a single 2D plane of a given sample at one time, so researchers can image samples at higher resolution. Compared with light-field microscopy, light-sheet microscopy produces images that are quicker to process, but the data are not as comprehensive, since they only capture information from a single 2D plane at a time.

To take advantage of the benefits of each technique, EMBL researchers developed an approach that uses light-field microscopy to image large 3D samples and light-sheet microscopy to train the AI algorithms, which then create an accurate 3D picture of the sample.

“If you build algorithms that produce an image, you need to check that these algorithms are constructing the right image,” explains Anna Kreshuk, the EMBL group leader whose team brought machine learning expertise to the project. In the new study, the researchers used light-sheet microscopy to make sure the AI algorithms were working, Anna says. “This makes our research stand out from what has been done in the past.”

Robert Prevedel, the EMBL group leader whose group contributed the novel hybrid microscopy platform, notes that the real bottleneck in building better microscopes often isn’t optics technology, but computation. That’s why, back in 2018, he and Anna decided to join forces. “Our method will be really key for people who want to study how brains compute. Our method can image an entire brain of a fish larva, in real time,” Robert says.

He and Anna say this approach could potentially be modified to work with different types of microscopes too, eventually allowing biologists to look at dozens of different specimens and see much more, much faster. For example, it could help to find genes that are involved in heart development, or could measure the activity of thousands of neurons at the same time.

Next, the researchers plan to explore whether the method can be applied to larger species, including mammals.

A paper describing the study, led by Penn State Professor of Biology Hong Ma, appears in the May 2021 issue of the journal Molecular Plant.

“Legumes make up the third largest family of flowering plants and are incredibly diverse — ranging from tiny herbs to giant trees,” said Ma, who is the Huck Distinguished Research Professor of Plant Molecular Biology at Penn State. “They are essential food crops for both humans and livestock, can be used as lumber, and have many other uses. Maybe most importantly, they can ‘fix’ nitrogen — extracting the vital nutrient from the atmosphere and storing it in nodules on their roots in a symbiotic relationship with soil bacteria — making them important as green manure to improve soil health.”

There are over 19,000 species in the legume family divided into six subfamilies and then further divided into narrower and narrower groupings based on their evolutionary relationships. There are 765 genera — the grouping one level above species — of which the team sampled members of 333. To build the family tree, the team analyzed gene sequences from the transcriptomes — the portion of the genome that is expressed as genes — of most of the 463 species and a small number of shallowly sequenced whole genomes from across legume diversity.

“This is the largest study of this kind for a single plant family,” said Ma. “We went to great lengths to sample as many species as we could to get a broad representation of the legume family, but it is often difficult to get well-preserved specimens that we can extract DNA or RNA from, especially for species found in remote locations. Having this broad representation of species allowed us to build the most detailed nuclear-gene family tree for legumes to date.”

In addition to helping researchers understand the evolution and diversification of legumes, the new legume family tree helps to clarify the relationship between crop plants and their wild relatives. Although the close relatives of important agricultural crops are often known, studying more distant wild cousins could reveal traits that could be exploited to help plants thrive in changing environments and resist diseases or insect pests.

Across the legume family tree, the research team identified strong evidence for 28 separate whole-genome duplication events. Whole-genome duplications, evolutionary events that result in complete duplication of the entire genome, are fairly common among flowering plants and are thought to allow for functional innovation and evolutionary diversification. One of the duplication events that the team identified appears to have occurred in the ancestor of all members of the legume family.

“Because for most of the species in our study we used transcriptomes and do not have entire genome sequences, we consider these as ‘proposed’ genome duplication events,” said Ma. “These kinds of studies are kind of like solving a mystery. If you only have one or a few witnesses it might be difficult to convince a jury of your evidence, but if you have a hundred witnesses who have different perspectives and they all point to the same thing it becomes difficult to dismiss that evidence. In our case, the different species are like our witnesses. The size of our study allowed us to identify events that we might otherwise have dismissed.”

The two largest subfamilies account for over 17,000 legume species and include all of the species with the ability to fix nitrogen. Nitrogen is an important plant nutrient — most commercial fertilizers contain a mix of nitrogen, phosphorus and potassium — so the symbiotic relationship between some legumes and the microorganisms that allow them to assimilate nitrogen from the atmosphere using root nodules has spurred their success by allowing them to colonize areas with less fertile soil. The research team also identified clues to the evolution of the genes responsible for this important trait.

“Our data support the idea that nodulation and nitrogen fixation originated a single time early in the history of legumes and other related nitrogen-fixing plants and the whole-genome duplication event at the origin of legumes might have been crucial for the evolution of this process,” said Ma. “In addition to this duplication event, we are also able to see gene loss in plants that do not have the ability to nodulate, and evolutionary changes in genes that contributed to their role in nodulation.”

Story Source:

Materials provided by Penn State. Original written by Sam Sholtis. Note: Content may be edited for style and length.

In their new analysis, published in the Journal of Clinical Sleep Medicine, the researchers found the number of medical visits and prescriptions filled were nearly doubled in people with sleep disorders such as sleep apnea and insomnia, compared to similar people without. Affected patients were also more likely to visit the emergency department and have more comorbid medical conditions.

Costly medical care for sleep disorder patients

The researchers sought out to determine the true diagnostic prevalence of sleep disorders and how expensive these conditions were to the health care system. They examined differences in health expenditures in similar patients with and without a sleep disorder diagnosis, as determined by their ICD-10 diagnosis code. The study included data from a nationally-representative survey of more than 22,000 Americans called the 2018 Medical Expenditure Panel Survey, which is administered by the Department of Health and Human Services’ Agency for Healthcare Research and Quality.

They found 5.6 percent of respondents had at least one sleep disorder, which translated to an estimated 13.6 million U.S. adults. This likely represents a significant underestimate, according to the authors, as insomnia alone is felt to conservatively affect 10 to 20 percent of the population. These individuals accumulated approximately $7,000 more in overall health care expenses per year compared to those without a sleep disorder — about 60 percent more in annual costs. This equates to a conservative estimate of $94.9 billion in health care costs per year attributable to sleep disorders.

The analysis revealed that patients with sleep disorders attended more than 16 office visits and nearly 40 medication prescriptions per year, compared to nearly 9 visits and 22 prescriptions for those without a sleep disorder. The study did not quantify non-health care related costs, but the authors noted it can be assumed that more doctors’ appointments means more time off from work, school or other social obligations, not to mention decreased productivity associated with symptoms, only exacerbating costs to society.

Sleep disorders raise risk for other conditions

Sleep disorders can take a toll on health and quality of life in numerous ways. Individuals with certain sleep disorders experience decrease daytime functionality related to sleepiness, mental fog and an increased risk of motor vehicle accidents, for instance. Obstructive sleep apnea is one of the most common sleep disorders and if untreated, can increase risk for neurocognitive issues, such as difficulty concentrating and mood disorders, as well as cardiovascular conditions including heart attacks, strokes, high blood pressure and irregular heart rhythms.

Getting a proper diagnosis at the sign of asleep problem can lead to an effective treatment for a sleep disorder.

“Fortunately, studies have demonstrated that treating certain sleep disorders effectively reduces health care utilization and costs. Therefore, sleep issues should not be ignored. Greater recognition of sleep disorders and an early referral to a sleep specialist are essential,” said Dr. Huyett. “Your sleep is important, and if there’s an issue with your sleep, seek help for it.”

Story Source:

Materials provided by Massachusetts Eye and Ear Infirmary. Note: Content may be edited for style and length.

The new species, named Ampelorhiza heteroxylon, belongs to a diverse group of tropical lianas called Paullinieae, within the soapberry family (Sapindaceae). More than 475 species of Paullinieae live in the tropics today.

Researchers identified the species from fossilized roots that revealed features known to be unique to the wood of modern climbing vines, adaptations that allow them to twist, grow and climb.

The study, “Climbing Since the Early Miocene: The Fossil Record of Paullinieae (Sapindaceae),” was published April 7 in the journal PLOS ONE.

“This is evidence that lianas have been creating unusual wood, even in their roots, as far back as 18 million years ago,” said wood anatomist Joyce Chery ’13, assistant research professor in the School of Integrative Plant Science, Plant Biology Section, in the College of Agriculture and Life Sciences, and a corresponding author of the paper.

“Before this discovery, we knew almost nothing about when or where these lianas evolved or how rapidly they diversified,” said first author Nathan Jud, assistant professor of plant biology at William Jewell College and a former Cornell postdoctoral researcher.

Panama was a peninsula 18.5 to 19 million years ago, a volcanic landscape covered with tropical forest in North America and separated from South America by a Central American seaway. While these forests contained North American animals, the plants mostly descended from South American tropical plants that had dispersed across the seaway, Jud said.

“The fossil we described is the oldest macrofossil of these vines,” he said, “and they were among the plants that made it to North America long before the Great American Biotic Interchange when large animals moved between the continents some 3 million years ago.”

In the study, the researchers made thin slices of the fossil, examined the arrangements and dimensions of tissues and water conducting vessels under a microscope and created a database of all the features. They then studied the literature to see how these features matched up with the living and fossil records of plants.

“We were able to say, it really does look like it’s a fossil from the Paullinieae group, given the anatomical characteristics that are similar to species that live today,” Chery said.

During their analyses, the researchers identified features that are characteristic of lianas. Most trees and shrubs have water-conducting tissues (which transport water and minerals from roots to leaves) that are all roughly the same size when viewed in a cross-section; in vines, these conduits come in two sizes, big and small, which is exactly what the researchers discovered in the fossil.

“This is a feature that is pretty specific to vines across all sorts of families,” Chery said.

The two vessel sizes provide insurance for a twisting and curving plant, as large vessels provide ample water flow, but are also vulnerable to collapse and develop cavities that disrupt flow. The series of smaller vessels offers a less vulnerable backup water transport system, Chery said.

Also, cross-sections of the wood in trees and shrubs are circular, but in the fossil, and in many living vines, such cross-sections are instead irregular and lobed.

Thirdly, on the walls of those vascular vessels, they found long horizontal perforations that allow for water to flow in lateral directions. That is a distinguishing feature of lianas in the soapberry family, Chery said.

In future work, now that they can place the lianas of Sapindaceae to 18.5 million years ago, the researchers intend to continue their investigation of the evolutionary history and diversification of this family. Chery also plans to investigate how wood has evolved in this group of vines, including identifying the genes that contribute to lobe-shaped stems.

The study was partly funded by the National Science Foundation.

University of Exeter scientists showed food pictures to more than 700 people, including omnivores (who eat meat and other foods), flexitarians (who try to eat less meat) and vegetarians.

About 7% of meat eaters (15% of flexitarians and 3% of omnivores) had a “fairly strong disgust response” to images of meat dishes commonly eaten in the UK, like roast chicken or bacon.

As a group, omnivores rated meat images about twice as disgusting on average as pictures of carbohydrate-rich foods like bread, chips and rice.

Based on the findings, the researchers say harnessing the “yuk factor” may be more effective than relying on willpower for anyone who wants to eat less meat.

“We were surprised to find that so many people are grossed out by meat — even people who eat meat all the time,” said Elisa Becker, of the University of Exeter.

“Our results don’t explain why these people eat meat, but it’s possible that habits, family and cultural traditions all play a part.

“Meat consumption is increasingly seen as unsustainable, unhealthy and unethical, and many people want to eat less meat.

“If you’re trying to cut down your meat intake, sheer willpower may not be enough — but harnessing the ‘yuk factor’ could be the way to go.”

The study’s 711 participants — 402 omnivores, 203 flexitarians and 106 vegetarians — each completed a survey and took a rapid-response task (measuring instinctive reactions) to test their levels of “meat disgust.”

“Meat liking” was also measured. About 75% of omnivores — and more than 20% of vegetarians — showed a fairly strong liking for meat.

To be classified as having “fairly strong” meat disgust, participants had to rate six meat images closer to “very much” than “not at all” on a sliding scale of disgust, and also had to show evidence of meat disgust on the rapid-response task.

Among flexitarians — the only group attempting to reduce their meat intake — meat disgust was a better predictor than self-control (measured in a separate questionnaire) of reduced meat-eating.

Meat disgust was also associated with reduced intake over the following six months.

“We hope that this information can help us develop new interventions to help people reduce their meat intake,” said Professor Natalia Lawrence, of the University of Exeter.

“Not everyone wants to reduce their meat consumption — but for those who do, we are working on computer tasks that might help them harness the power of disgust in a fun way.

“It’s important to note that our study does not establish causation — so further research is needed to find out whether meat disgust causes people to eat less meat, or whether avoiding meat allows these negative emotional responses to develop or be expressed.”

Becker added: “It’s interesting to note that almost all of us experience meat disgust from time to time — for example when we see unfamiliar meats or dishes made from parts of animals we don’t usually eat, like squirrel meat or beef heart.

“Humans may have evolved a degree of meat disgust because eating spoilt meat can be much more dangerous than eating a carrot that’s a bit off.”

Story Source:

Materials provided by University of Exeter. Note: Content may be edited for style and length.

Most of the many millions of songbirds that migrate every year between Europe and Africa fly by night and spend the daytime hours resting and eating. Some species, which normally only fly by night, occasionally fly for over 24 consecutive hours to avoid having to stop in inhospitable locations with insufficient access to food, such as deserts and seas. The great reed warbler, Acrocephalus arundinaceus, is such a species. During its month-long migration, it can fly for up to 34 consecutive hours without landing.

A research team from Lund University in Sweden, University of Copenhagen and the Nature Research Centre in Vilnius has used unique data loggers developed at Lund University to study these long migratory flights almost minute by minute. The researchers used the data loggers, weighing no more than 1.2 grams, to track 14 great reed warblers which were captured during the summer at Lake Kvismaren, close to Örebro in Central Sweden.

The data loggers collected data over a full year, thereby providing the researchers with valuable information. As the great reed warblers migrated to and from Africa, the data loggers continuously stored information about altitude, activity and the sun’s ascend and descend. The altitude measurements were recorded once per hour and every five minutes the data loggers saved information about whether the birds were flying, looking for food on the ground or resting.

Based on this information, the researchers calculated that the great reed warblers flew at an average altitude of 2400 metres at night. On the few occasions on which the birds flew for more than one consecutive night, they climbed 3000 metres at dawn to fly at an average altitude of 5400 metres during the day. They remained cruising at that altitude until dusk, when they dived 3-4000 metres and continued their flight at an altitude of around 2000 metres the following night.

Sissel Sjöberg and Dennis Hasselquist at Lund University led the study, and the results came as a surprise to them and their colleagues. They describe the great reed warblers’ behaviour as very consistent, indicating that the birds may have to fly at more than twice their night-time altitude to manage migrating during daylight hours at all.

There is no definitive explanation as yet. However, the research team rules out the previously predominant explanations for choice of altitudes in migratory birds: winds and air temperature.

“At altitudes over 2000 metres, winds and air temperatures do not change merely going from day to night or vice versa,” says Sissel Sjöberg.

Nor are there any high mountains along the migration route that could push the birds up to extreme altitudes. Two possible explanations could be that during daytime, the great reed warblers see further from a great height and that they can avoid birds of prey. But the researchers also suggest another explanation: the birds climb to more than double the altitude during the day to avoid becoming overheated when they are exposed to the additional effect of the sun’s heat radiation.

“It is colder higher up in the air and when the birds rise to 5400 metres, they reach a layer of air with a temperature of around minus 9 degrees centigrade. That is 22 degrees colder than the altitude at which the birds fly during the night,” says Sissel Sjöberg.

“The wings of migratory songbirds beat several times per second, so they are working extremely hard, which makes their bodies very warm regardless of whether they are flying by day or by night. But if they fly by day, they are also exposed to the heat from the sun’s radiation; that is what we suggest they are compensating for when they climb to a much colder layer of air at daytime. It is likely that they would not manage to fly by day without becoming overheated if they did not climb to these extreme altitudes,” says Dennis Hasselquist.

The researchers are continuing to study the great reed warblers. In time, they hope to find answers to the question of why the birds behave in this way, which might also help to explain why most birds that migrate to the tropics during the winter fly almost entirely by night, whereas they are active during the day and sleep at night for the rest of the year.

Story Source:

Materials provided by Lund University. Note: Content may be edited for style and length.

Two papers published in The Astrophysical Journal report these findings, with Kyle Boone as lead author. Currently a postdoctoral fellow at the University of Washington, Boone is a former graduate student of Nobel Laureate Saul Perlmutter, the Berkeley Lab senior scientist and UC Berkeley professor who led one of the teams that originally discovered dark energy. Perlmutter was also a co-author on both studies.

Supernovae were used in 1998 to make the startling discovery that the expansion of the universe is speeding up, rather than slowing down as had been expected. This acceleration — attributed to the dark energy that makes up two-thirds of all the energy in the universe — has since been confirmed by a variety of independent techniques as well as with more detailed studies of supernovae.

The discovery of dark energy relied on using a particular class of supernovae, Type Ia. These supernovae always explode with nearly the same intrinsic maximum brightness. Because the observed maximum brightness of the supernova is used to infer its distance, the small remaining variations in the intrinsic maximum brightness limited the precision with which dark energy could be tested. Despite 20 years of improvements by many groups, supernovae studies of dark energy have until now remained limited by these variations.

Quadrupling the number of supernovae

The new results announced by the SNfactory come from a multi-year study devoted entirely to increasing the precision of cosmological measurements made with supernovae. Measurement of dark energy requires comparisons of the maximum brightnesses of distant supernovae billions of light-years away with those of nearby supernovae “only” 300 million light-years away. The team studied hundreds of such nearby supernovae in exquisite detail. Each supernova was measured a number of times, at intervals of a few days. Each measurement examined the spectrum of the supernova, recording its intensity across the wavelength range of visible light. An instrument custom-made for this investigation, the SuperNova Integral Field Spectrometer, installed at the University of Hawaii 2.2-meter telescope at Maunakea, was used to measure the spectra.

“We’ve long had this idea that if the physics of the explosion of two supernovae were the same, their maximum brightnesses would be the same. Using the Nearby Supernova Factory spectra as a kind of CAT scan through the supernova explosion, we could test this idea,” said Perlmutter.

Indeed, several years ago, physicist Hannah Fakhouri, then a graduate student working with Perlmutter, made a discovery key to today’s results. Looking at a multitude of spectra taken by the SNfactory, she found that in quite a number of instances, the spectra from two different supernovae looked very nearly identical. Among the 50 or so supernovae, some were virtually identical twins. When the wiggly spectra of a pair of twins were superimposed, to the eye there was just a single track. The current analysis builds on this observation to model the behavior of supernovae in the period near the time of their maximum brightness.

The new work nearly quadruples the number of supernovae used in the analysis. This made the sample large enough to apply machine-learning techniques to identify these twins, leading to the discovery that Type Ia supernova spectra vary in only three ways. The intrinsic brightnesses of the supernovae also depend primarily on these three observed differences, making it possible to measure supernova distances to the remarkable accuracy of about 3%.

Just as important, this new method does not suffer from the biases that have beset previous methods, seen when comparing supernovae found in different types of galaxies. Since nearby galaxies are somewhat different than distant ones, there was a serious concern that such dependence would produce false readings in the dark energy measurement. Now this concern can be greatly reduced by measuring distant supernovae with this new technique.

In describing this work, Boone noted, “Conventional measurement of supernova distances uses light curves — images taken in several colors as a supernova brightens and fades. Instead, we used a spectrum of each supernova. These are so much more detailed, and with machine-learning techniques it then became possible to discern the complex behavior that was key to measuring more accurate distances.”

The results from Boone’s papers will benefit two upcoming major experiments. The first experiment will be at the 8.4-meter Rubin Observatory, under construction in Chile, with its Legacy Survey of Space and Time, a joint project of the Department of Energy and the National Science Foundation. The second is NASA’s forthcoming Nancy Grace Roman Space Telescope. These telescopes will measure thousands of supernovae to further improve the measurement of dark energy. They will be able to compare their results with measurements made using complementary techniques.

Aldering, also a co-author on the papers, observed that “not only is this distance measurement technique more accurate, it only requires a single spectrum, taken when a supernova is brightest and thus easiest to observe — a game changer!” Having a variety of techniques is particularly valuable in this field where preconceptions have turned out to be wrong and the need for independent verification is high.

The SNfactory collaboration includes Berkeley Lab, the Laboratory for Nuclear Physics and High Energy at Sorbonne University, the Center for Astronomical Research of Lyon, the Institute of Physics of the 2 Infinities at the University Claude Bernard, Yale University, Germany’s Humboldt University, the Max Planck Institute for Astrophysics, China’s Tsinghua University, the Center for Particle Physics of Marseille, and Clermont Auvergne University.

This work was supported by the Department of Energy’s Office of Science, NASA’s Astrophysics Division, the Gordon and Betty Moore Foundation, the French National Institute of Nuclear and Particle Physics and the National Institute for Earth Sciences and Astronomy of the French National Centre for Scientific Research, the German Research Foundation and German Aerospace Center, the European Research Council, Tsinghua University, and the National Natural Science Foundation of China.

Additional background

In 1998, two competing groups studying supernovae, the Supernova Cosmology Project and the High-z Supernova Search team, both announced they had found evidence that, contrary to expectations, the expansion of the universe was not slowing but becoming faster and faster. Dark energy is the term used to describe the cause of the acceleration. The 2011 Nobel Prize was awarded to leaders of the two teams: Saul Perlmutter of Berkeley Lab and UC Berkeley, leader of the Supernova Cosmology Project, and to Brian Schmidt of the Australian National University and Adam Riess of Johns Hopkins University, from the High-z team.

Additional techniques for measuring dark energy include the DOE-supported Dark Energy Spectroscopic Instrument, led by Berkeley Lab, which will use spectroscopy on 30 million galaxies in a technique called baryon acoustic oscillation. The Rubin Observatory will also use another called weak gravitational lensing.

Informatics experts at the University of Sussex have developed a new computational model that demonstrates a long sought after link between insect and mammalian learning, as detailed in a new paper published today in Nature Communications.

Incorporating anatomical and functional data from recent experiments, Dr James Bennett and colleagues modelled how the anatomy and physiology of the fruit fly’s brain can support learning according to the reward prediction error (RPE) hypothesis.

The computational model indicates how dopamine neurons in an area of a fruit fly’s brain, known as the mushroom body, can produce similar signals to dopamine neurons in mammals, and how these dopamine signals can reliably instruct learning.

The academics believe that establishing whether flies also use prediction errors to learn could lead to more humane animal research allowing researchers to replace animals with more simple insect species for future studies into the mechanisms of learning.

By opening up new opportunities to study neural mechanisms of learning, the researchers hope the model could also be helpful in illuminating greater understanding of mental health issues such as depression or addiction which are underpinned by the RPE hypothesis.

Dr Bennett, research fellow in the University of Sussex’s School of Engineering and Informatics, said: “Using our computational model, we were able to show that data from insect experiments did not necessarily conflict with predictions from the RPE hypothesis, as had been thought previously.

“Establishing a bridge between insect and mammal studies on learning may open up the possibility to exploit the powerful genetic tools available for performing experiments in insects, and the smaller scale of their brains, to make sense of brain function and disease in mammals, including humans.”

Understanding of how mammals learn has come a long way thanks to the RPE hypothesis, which suggests that associative memories are learned in proportion to how inaccurate they are.

The hypothesis has had considerable success explaining experimental data about learning in mammals, and has been extensively applied to decision-making and mental health illnesses such as addiction and depression. But scientists have encountered difficulties when applying the hypothesis to learning in insects due to conflicting results from different experiments.

The University of Sussex research team created a computational model to show how the major features of mushroom body anatomy and physiology can implement learning according to the RPE hypothesis.

The model simulates a simplification of the mushroom body, including different neuron types and the connections between them, and how the activity of those neurons promote learning and influence the decisions a fly makes when certain choices are rewarded.

To further understanding of learning in fly brains, the research team used their model to make five novel predictions about the influence different neurons in the mushroom body have on learning and decision-making, in the hope that they promote future experimental work.

Dr Bennett said: “While other models of the mushroom body have been created, to the best of our knowledge no other model until now has included connections between dopamine neurons and another set of neurons that predict and drive behaviour towards rewards. For example, when the reward is the sugar content of food, these connections would allow the predicted sugar availability to be compared with the actual sugar ingested, allowing more accurate predictions and appropriate sugar-seeking behaviours to be learned.

“The model can explain a large array of behaviours exhibited by fruit flies when the activity of particular neurons in their brains are either silenced or activated artificially in experiments. We also propose connections between dopamine neurons and other neurons in the mushroom body, which have not yet been reported in experiments, but would help to explain even more experimental data.”

Thomas Nowotny, Professor of Informatics at the University of Sussex, said: “The model brings together learning theory and experimental knowledge in a way that allows us to think systematically how fly brains actually work. The results show how learning in simple flies might be more similar to how we learn than previously thought.”

Story Source:

Materials provided by University of Sussex. Original written by Neil Vowles. Note: Content may be edited for style and length.

“One of the things that is foremost in the minds of anesthesiologists is ‘Do I have somebody who is lying in front of me who may be conscious and I don’t realize it?’ Being able to reliably maintain unconsciousness in a patient during surgery is fundamental to what we do,” said senior author Emery N. Brown, Edward Hood Taplin Professor in The Picower Institute for Learning and Memory and the Institute for Medical Engineering and Science at MIT, and an anesthesiologist at MGH. “This is an important step forward.”

More than providing a good readout of unconsciousness, Brown added, the new algorithms offer the potential to allow anesthesiologists to maintain it at the desired level while using less drug than they might administer when depending on less direct, accurate and reliable indicators. That can improve patient’s post-operative outcomes, such as delirium.

“We may always have to be a little bit ‘overboard’,” said Brown, who is also a professor at Harvard Medical School. “But can we do it with sufficient accuracy so that we are not dosing people more than is needed?”

Used to drive an infusion pump, for instance, algorithms could help anesthesiologists precisely throttle drug delivery to optimize a patient’s state and the doses they are receiving.

Artificial intelligence, real-world testing

To develop the technology to do so, postdocs John Abel and Marcus Badgeley led the study, published in PLOS ONE [LINK TBD], in which they trained machine learning algorithms on a remarkable data set the lab gathered back in 2013. In that study, 10 healthy volunteers in their 20s underwent anesthesia with the commonly used drug propofol. As the dose was methodically raised using computer controlled delivery, the volunteers were asked to respond to a simple request until they couldn’t anymore. Then when they were brought back to consciousness as the dose was later lessened, they became able to respond again. All the while, neural rhythms reflecting their brain activity were recorded with electroencephalogram (EEG) electrodes, providing a direct, real-time link between measured brain activity and exhibited unconsciousness.

In the new work, Abel, Badgeley and the team trained versions of their AI algorithms, based on different underlying statistical methods, on more than 33,000 two-second-long snippets of EEG recordings from seven of the volunteers. This way the algorithms could “learn” the difference between EEG readings predictive of consciousness and unconsciousness under propofol. Then the researchers tested the algorithms in three ways.

First, they checked whether their three most promising algorithms accurately predicted unconsciousness when applied to EEG activity recorded from the other three volunteers of the 2013 study. They did.

Then they used the algorithms to analyze EEG recorded from 27 real surgery patients who received propofol for general anesthesia. Even though the algorithms were now being applied to data gathered from a “noisier” real-world surgical setting where the rhythms were also being measured with different equipment, the algorithms still distinguished unconsciousness with higher accuracy than other studies have shown. The authors even highlight one case in which the algorithms were able to detect a patient’s decreasing level of unconsciousness several minutes before the actual attending anesthesiologist did, meaning that if it had been in use during the surgery itself, it could have provided an accurate and helpful early warning.

As a third test, the team applied the algorithms to EEG recordings from 17 surgery patients who were anesthetized with sevoflurane. Though sevoflurane is different from propofol and is inhaled rather than infused, it works in a similar manner, by binding to the same GABA-A receptors on the same key types of brain cells. The team’s algorithms again performed with high, though somewhat reduced accuracy, suggesting that their ability to classify unconsciousness carried over reliably to another anesthetic drug that works in a similar way.

The ability to predict unconsciousness across different drugs with the same mechanism of action is key, the authors said. One of the main flaws with current EEG-based systems for monitoring consciousness, they said, is that they don’t distinguish among drug classes, even though different categories of anesthesia drugs work in very different ways, producing distinct EEG patterns. They also don’t adequately account for known age differences in brain response to anesthesia. These limitations on their accuracy have also limited their clinical use.

In the new study, while the algorithms trained on 20-somethings applied well to cohorts of surgery patients whose average age skewed significantly older and varied more widely, the authors acknowledge that they want to train algorithms distinctly for use with children or seniors. They can also train new algorithms to apply specifically for other kinds of drugs with different mechanisms of action. All together a suite of well trained and attuned algorithms could provide high accuracy that accounts for patient age and the drug in use.

Abel said the team’s approach of framing the problem as a matter of predicting consciousness via EEG for a specific class of drugs made the machine learning approach very simple to implement and extend.

“This is a proof of concept showing that now we can go and say let’s look at an older population or let’s look at a different kind of drug,” he said. “Doing this is simple if you set it up the right way.”

The resulting algorithms aren’t even computationally demanding. The authors noted that for a given 2 seconds of EEG data, the algorithms could make an accurate prediction of consciousness in less than a tenth of a second running on just a standard MacBook Pro computer.

The lab is already building on the findings to refine the algorithms further, Brown said. He said he also wants to expand testing to hundreds more cases to further confirm their performance, and also to determine whether wider distinctions may begin to emerge among the different underlying statistical models the team employed.