Catbas collaborated with his former civil engineering student Marwan Debees ’23PhD, who now works as a NASA Bridge Program manager, on newly published research that details how infrared thermography, high-definition imaging and neural network analysis can combine to make concrete bridge inspections more efficient.

Catbas and Debees are hopeful that their findings, recently published in the Transportation Research Record, can be leveraged by engineers through a combination of these methods to strategically pinpoint bridge conditions and better allocate repair costs.

“If we better understand which bridges need more repairs and which bridges may be postponed, then [funding agencies] can use limited funds more wisely, and then we can direct our efforts to the really critical bridges,” Catbas says. “We have about 650,000 bridges in the U.S. and we have been working to examine how we can use novel technologies to understand the existing condition of structures.”

Debees noted an instance during a NASA bridge load test where Catbas and his team assisted in evaluating the repairs. They determined that the repairs made were sufficient, ultimately, eliminating the next phrase of planned work.

“We’re only spending the money where we need to instead of doing it without a comprehensive understanding of the actual conditions of the bridge in the field,” Debees says. “The goal is to better understand the conditions of the bridge and have a better priority list of what bridges are really in need.”

Diagnosing Concrete Bridges

Catbas says what he and other civil engineers do to assess a structure’s overall integrity may be likened to a doctor’s diagnostics for a person’s wellbeing.

“Structural health monitoring, which is almost like human health monitoring, is where we use different types of equipment to better understand the safety and serviceability of structures,” he says.

To help take high-definition images to compare to infrared data, the researchers closely collaborated with NEXCO-West USA. Inc, an imaging and non-destructive evaluation company in Tysons, Virginia, that have specialized vehicles equipped with imaging tools. With the company’s support, the research team utilized the infrared data to assess the conditions of bridge components, including the deck, superstructure and substructure.

“As far as the infrared itself, there are some limitations,” Debees says. “One of the things in this paper that helped overcome some of these limitations is high-definition images to complement the infrared images.”

These technologies that were used in the study by Catbas and Debees provided a more comprehensive record of concrete bridge health.

“Human visualization has limitations,” Catbas says. “It’s almost like a doctor just looking at you and saying that you look fine when you might really be fine, or you might not be. There may be other problems that the sensors and other technologies can tell you, kind of like when a doctor says he wants more testing, so he sends you to get an X-ray or an MRI. We are taking a similar approach to our bridges.”

Bridging the Gap Between Technology and Interpretation

Infrared thermography works by collecting a structure’s thermal responses, which can indicate defects within it such as heat loss, moisture intrusion or other structural problems.

To analyze the different parts of the bridge such as the deck, superstructure and substructure, the research team used thermography and image capturing technologies deployed on boats under the bridge and on vehicles traveling across it so that traffic wouldn’t be impeded and motorists may continue using the roads.

The combination of visual inspection and imaging is common practice, but Debees says the element of utilizing a neural network and machine learning to decipher the data is something that is an emerging component of inspections. The collective knowledge from experienced engineers doing similar inspections was used to compare the results in the study.

“The way it differs from other utilization is that we are not using just infrared cameras and collecting raw data, but then we have a level of post-processing, and we are eliminating the noise or unnecessary information within the infrared image,” Debees says. “Then we use this data to understand where these defects are and then we integrate them within the current required bridge inspection processes. We close the loop by using some decision-making and algorithms with an easy-to-use perceptron neural network to guide the inspector or engineer without spending too much time or data analysis.”

The two parts of the paper are how to implement this new technology and how it can be used to accelerate decision making while keeping it accurate and safe, he says.

“When we do bridge inspections, we aim to find ways to accelerate or make it more efficient while also having more data to rely on in the future or in the immediate decision making,” Debees says. “We can determine which bridge needs to be evaluated right away, which needs more testing and we can see the significance of the finding quicker.”

Crossing Into the Future

Debees says one of the most exciting parts of the research findings is the realization that the framework of multiple inspection techniques can be integrated with collective knowledge and applied to monitor a wide variety of structures.

“We’re not limited to concrete bridges,” he says. “We can build on this research and applying it with different inspection methods and use it for different infrastructure types. We can try this on concrete buildings, or steel bridges, buildings or other structures.”

Using machine learning and collective knowledge to interpret data is something that Debees believes will continue to have a role in inspections even beyond the purview of their study.

“I think what was eye-opening to me is there is room, even outside of conventional inspections, to utilize more decision-making neural networks to standardize the decision-making [process],” he says. “You can make it easier on the people in the field to know where to make decisions on the spot or where to seek more experienced help.”

There are ample opportunities to discover even more innovative ways to assess structural health, and Catbas says he gladly looks forward to meeting the next challenge with former students and collaborators like Debees.

“Like other Ph.D. students of mine, we still keep in touch once they graduate and then become my colleague,” Catbas says as he turns to Debees. “So, my question is this: ‘What are we going to work on next?'”

In a study spanning meerkats in the Kalahari desert, coatis in Panama’s rainforest, and spotted hyenas in Kenya’s savanna, researchers have discovered that the daily actions of these animals show surprisingly similar patterns. Whether a meerkat scratches in the sand for scorpions or a coati rests in the canopy, a shared ordering of the behaviors persists across different landscapes, species, individuals, and types of behaviors. To the international team of fourteen authors, led by researchers at the Max Planck Institute of Animal Behavior, the findings are unexpected and — possibly — profound.

“We assumed there would be differences,” said Pranav Minasandra, a postdoctoral researcher at MPI-AB and lead author of the study in PNAS. After all, differences are apparent when comparing meerkats, coatis, and hyenas, which occupy dissimilar environments and ecological roles. “But we found common patterns in how animals switch between behaviors, regardless of what species and which individual. It’s as if their behavior was built on the same hidden algorithm.”

Uncovering underlying patterns

The hidden algorithm came to light in data that were collected from wild animals tagged with accelerometers — the same small sensors in phones and watches that track our activity. The species studied are all social mammals, but they differ in their ecology and behavior. Spotted hyenas are large carnivores, meerkats are small burrowing animals, and coatis are racoon-sized tree-dwellers. Accelerometers measure posture changes many times each second and the recordings can continue for several days. These high-resolution motion traces collected from animals were then classified using machine learning into behavioral states like lying, foraging, and walking. For instance, a meerkat might lie down for 10 minutes then briefly stand up to look around for 20 seconds before moving around to search for food for another few minutes.

“This approach allowed us to capture detailed behavioral sequences over days and even weeks from multiple individuals across three distinct species,” says Ariana Strandburg-Peshkin, group leader at MPI-AB and senior author on the study.

Across behaviors, individuals, and species, one common principle emerged: the longer an animal stays in one behavioral state, the less likely it is to change it in the next moment. “This was unexpected,” adds Minasandra.

Imagine a hyena walking continuously for 10 minutes. Most people would probably guess that the hyena would be more likely to stop over time, and the authors did too. “We originally thought the probability of switching behaviors would increase over time, as we assumed it would not be optimal to lock-in to any behavior.” Remarkably, this kind of lock-in, also called a decreasing hazard function, was consistent across all studied animals and species.

The authors further examined how current behavior predicts future actions — a concept they call “predictivity decay.” Predictivity decay reflects the increasing difficulty in predicting behavior the further we look into the future, primarily due to random, unpredictable variations. The shape of the decay graph conveys how decision-making systems across different timescales interact to generate animals’ behavioral sequences. “We found that the pattern of predictivity decay was remarkably consistent across all animals studied, implying a shared architecture beneath the surface.”

Why these patterns?

The study raises a big question: Why do such patterns occur? The authors propose two broad explanations.

First is positive feedback: the longer an animal remains in a state — say, lying down — the more likely that staying put is rewarded, whether because it’s warm, safe, or socially reinforced. Behavior becomes self-reinforcing.

The second possibility is multi-timescale decision-making. Instead of a single internal clock governing when to switch behaviors, animals may integrate cues from many processes — internal hunger, external threats, social context — each with its own tempo. The interplay of these overlapping signals could generate the observed patterns.

Future studies may explore whether these patterns hold in other animals beyond the three mammals in the study: non-social species, across developmental stages, or under different ecological pressures. There’s also the question of whether these long-time behaviors offer advantages — perhaps by optimizing attention, conserving energy, or enhancing group coordination.

Says co-author Meg Crofoot, Director of the Department for the Ecology of Animal Societies: “What this study suggests is that real animals, be they hunting, hiding, or resting, are guided by hidden structures that seem to echo across life’s branches.”

“Methane is a very potent greenhouse gas that we need to address urgently,” said co-author Xin (Lindsay) Lan, a climate scientist at CU Boulder’s Cooperative Institute for Research in Environmental Sciences (CIRES). “Our study suggests that a significant portion of the recent rise in atmospheric methane originates from natural sources driven by climate change. Our emission reduction efforts need to be more aggressive.”

Methane is the second most abundant human-produced greenhouse gas after carbon dioxide. But an equal amount of methane traps about 30 times more heat than CO₂ over a 100-year time frame. Methane has been responsible for roughly a quarter of the planet’s warming since the Industrial Revolution.

Lan has spent the past decade tracking methane concentrations in the atmosphere at Boulder’s Global Monitoring Laboratory at the National Oceanic and Atmospheric Administration (NOAA).

Lan and her colleagues at NOAA have observed a rapid increase in atmospheric methane levels in recent years. While previous studies have shown fossil fuel production accounts for 30% of global methane emissions, Lan and colleagues have noticed a steady increase in emissions from microbial sources since 2007.

These microbes, specifically a group known as archaea, produce methane as a byproduct of their metabolism in environments like wetlands, landfills and livestock’s digestive systems.

Together, microbial emissions contribute to nearly half of global methane emissions, but it remains unclear which specific sources are driving this increase.

“While long-term methane trends are important to investigate, we also need to look at seasonal variations to understand how individual sources are changing and how the natural mechanisms that remove methane from the atmosphere are evolving,” Lan said.

A vicious cycle

To get a clearer picture, Lan and her team analyzed seasonal fluctuations in atmospheric methane levels over the past four decades.

They found that methane’s seasonal amplitude — the difference between peak and lowest methane levels within a year — has been decreasing in northern high-latitude regions, including the Arctic.

Using computer models, the team showed that this trend since the 1980s is largely a result of increased methane emissions from wetlands. Increased precipitation in the Arctic has expanded the region’s wetlands by 25% during the warmer months. Rising temperatures have also been melting some of the perpetually frozen soil layer deep underground, known as permafrost, in summer.

The melted, waterlogged soils have provided ideal conditions for archaea to thrive, leading to higher methane emissions which in turn could accelerate warming further.

Scientists have long warned about such climate feedback loops, but the precise scale and speed of these effects remain uncertain. Lan said this new study added another piece of evidence that natural methane emissions have already been responding to a warming climate.

“This study, along with a few previous studies, has provided indirect evidence on potential climate feedback on methane emissions, which would be beyond our ability to control directly,” Lan said.

The sharp increase in atmospheric methane and its climate feedback effects since 2007 resemble the planet’s most dramatic warming events that brought past ice ages to an end, according to Lan’s previous research.

“Our hope is that by rapidly reducing emissions, we can avoid triggering more severe and abrupt climate feedback that could lead to catastrophic events,” she said.

Methane sponges

The team’s simulations also found a 10% increase in the levels of hydroxyl (OH) radical since 1984. These radicals are highly reactive molecules that can soak up and remove methane and other air pollutants.

Because these molecules stay in the air for less than a second before they react with other compounds, scientists cannot directly measure them globally. In the past, researchers had assumed the OH levels remained constant over the years when calculating atmospheric methane emissions, but this study suggested that assumption might be wrong.

“Our result showed that we’ve been underestimating how much methane the atmosphere has been removing, which means that there’s actually more methane being emitted than we previously estimated,” Lan said.

Understanding the specific source of emission is vital in designing climate mitigation policies. While microbial emissions are responsible for most of the methane growth, human-produced methane from burning fossil fuels remains an important contributor.

“We need to aggressively cut all greenhouse gas emissions from the sources we can control,” Lan said. She added that the world’s permafrost currently holds at least twice as much carbon as is currently in the atmosphere. If future warming causes widespread permafrost thaw and releases that carbon, it could trigger irreversible changes to the planet’s climate. “We need to address the feedback loop before reaching that tipping point.”

Humans perceive and navigate the world around us with the help of our five senses: sight, hearing, touch, taste and smell. And while scientists have long known that these different senses activate different parts of the brain, a new Yale-led study indicates that multiple senses all stimulate a critical region deep in the brain that controls consciousness.

The study, published May 15 in the journal NeuroImage, sheds new light on how sensory perception works in the brain and may fuel the development of therapies to treat disorders involving attention, arousal, and consciousness

In the study, a research team led by Yale’s Aya Khalaf focused on the workings of subcortical arousal systems, brain structure networks that play a crucial role in regulating sleep-wake states. Previous studies on patients with disorders of consciousness — such as coma or epilepsy — have confirmed the influence of these systems on states of consciousness.

But prior research has been largely limited to tracking individual senses. For the new study, researchers asked if stimuli from multiple senses share the same subcortical arousal networks. They also looked at how shifts in a subject’s attention might affect these networks.

For the study, researchers analyzed fMRI (functional magnetic resonance imaging) datasets collected from 1,561 healthy adult participants as they performed 11 different tasks using four senses: vision, audition, taste, and touch.

They made two important discoveries: that sensory input does make use of shared subcortical systems and, more surprisingly, that all input — regardless of which sense delivered the signal — stimulates activity in two deep brain regions, the midbrain reticular formation and the central thalamus, when a subject is sharply focused on the senses.

The key to stimulating the critical central brain regions, they found, were the sudden shifts in attention demanded by the tasks.

“We were expecting to find activity on shared networks, but when we saw all the senses light up the same central brain regions while a test subject was focusing, it was really astonishing,” said Khalaf, a postdoctoral associate in neurology at Yale School of Medicine and lead author of the study.

The discovery highlighted how key these central brain regions are in regulating not only disorders of consciousness, but also conditions that impact attention and focus, such as attention deficit hyperactivity disorder. This finding could lead to better targeted medications and brain stimulation techniques for patients.

“This has also given us insights into how things work normally in the brain,” said senior author Hal Blumenfeld, the Mark Loughridge and Michele Williams Professor of Neurology who is also a professor in neuroscience and neurosurgery and director of the Yale Clinical Neuroscience Imaging Center. “It’s really a step forward in our understanding of awareness and consciousness.”

Looking across senses, this is the first time researchers have seen a result like this, said Khalaf, who is also part of Blumenfeld’s lab.

“It tells us how important this brain region is and what it could mean in efforts to restore consciousness,” she said.

Other authors include Erick Lopez, a former undergraduate researcher in Blumenfeld’s lab, and collaborators from Harvard Medical School.

This research was supported in part by funding from the National Institutes of Health.

In spring 2011, Joplin, Missouri, was devastated by an EF5 tornado with estimated winds exceeding 200 mph. The storm caused 161 fatalities, injured over 1,000 people, and damaged and destroyed around 8,000 homes and businesses. The tornado carved a mile-wide path through the densely populated south-central area of the city, leaving behind miles of splintered rubble and causing over $2 billion in damage.

The powerful winds of tornadoes often surpass the design limits of most residential and commercial buildings. Traditional methods of assessing damage after a disaster can take weeks or even months, delaying emergency response, insurance claims and long-term rebuilding efforts.

New research from Texas A&M University might change that. Led by Dr. Maria Koliou, associate professor and Zachry Career Development Professor II in the Zachry Department of Civil and Environmental Engineering at Texas A&M, researchers have developed a new method that combines remote sensing, deep learning and restoration models to speed up building damage assessments and predict recovery times after a tornado. Once post-event images are available, the model can produce damage assessments and recovery forecasts in less than an hour.

The researchers published their model in Sustainable Cities and Society.

“Manual field inspections are labor-intensive and time-consuming, often delaying critical response efforts,” said Abdullah Braik, coauthor and a civil engineering doctoral student at Texas A&M. “Our method uses high-resolution sensing imagery and deep learning algorithms to generate damage assessments within hours, immediately providing first responders and policymakers with actionable intelligence.”

The model does more than assess damage — it also helps predict repair costs and estimate recovery times. Researchers can assess these timelines and costs in different situations by combining deep learning technology, a type of artificial intelligence, with advanced recovery models.

“We aim to provide decision-makers with near-instantaneous damage assessment and probabilistic recovery forecasts, ensuring that resources are allocated efficiently and equitably, particularly for the most vulnerable communities,” Braik said. “This enables proactive decision-making in the aftermath of a disaster.”

How It Works

Researchers combined three tools to create the model: remote sensing, deep learning and restoration modeling.

Remote sensing uses high-resolution satellite or aerial images from sources such as NOAA to show the extent of damage across large areas.

“These images are crucial because they offer a macro-scale view of the affected area, allowing for rapid, large-scale damage detection,” Braik said.

Deep learning automatically analyzes these images to identify the severity of the damage accurately. The AI is trained before disasters by analyzing thousands of images of past events, learning to recognize visible signs of damage such as collapsed roofs, missing walls and scattered debris. The model then classifies each building into categories such as no damage, moderate damage, major damage, or destroyed.

Restoration modeling uses past recovery data, building and infrastructure details and community factors — like income levels or access to resources — to estimate how long it might take for homes and neighborhoods to recover under different funding or policy conditions.

When these three tools are combined, the model can quickly assess the damage and predict short- and long-term recovery timelines for communities affected by disasters.

“Ultimately, this research bridges the gap between rapid disaster assessment and strategic long-term recovery planning, offering a risk-informed yet practical framework for enhancing post-tornado resilience,” Braik said.

Testing The Model

Koliou and Braik used data from the 2011 Joplin tornado to test their model due to its massive size, intensity and availability of high-quality post-disaster information. The tornado destroyed thousands of buildings, creating a diverse dataset that allowed the model to be trained and tested across various levels of structural damage. Detailed ground-level damage assessments provided a reliable benchmark to check how accurately the model could classify the severity of the damage.

“One of the most interesting findings was that, in addition to detecting damage with high accuracy, we could also estimate the tornado’s track,” Braik said. “By analyzing the damage data, we could reconstruct the tornado’s path, which closely matched the historical records, offering valuable information about the event itself.”

Future Directions

Researchers are working on using this model for other types of disasters, such as hurricanes and earthquakes, as long as satellites can detect damage patterns.

“The key to the model’s generalizability lies in training it to use past images from specific hazards, allowing it to learn the unique damage patterns associated with each event,” Braik said. “We have already tested the model on hurricane data, and the results have shown promising potential for adapting to other hazards.”

The research team believes their model could be critical in future disaster response, helping communities recover faster and more efficiently. The team wants to extend the model beyond damage assessment to include real-time updates on recovery progress and tracking recovery over time.

“This will allow for more dynamic and informed decision-making as communities rebuild,” he said. “We aim to create a reliable tool that enhances disaster management efficiency and supports quicker recovery efforts.”

The technology has the potential to transform how emergency officials, insurers and policymakers respond in the crucial hours and days after a storm by delivering near-instant assessments and recovery projections.

Funding for this research was provided by the National Science Foundation.

University of Oklahoma researcher Matthew Potthoff, Ph.D., is the lead author of the study, which provides valuable insight about the mechanism of action of the hormone, which is a target for a new class of highly anticipated drugs that are in Phase 3 clinical trials.

“Fatty liver disease, or MASLD (metabolic dysfunction-associated steatotic liver disease), is a buildup of fat in the liver. It can progress to MASH (metabolic dysfunction-associated steatohepatitis) during which fibrosis and, ultimately, cirrhosis can occur. MASLD is becoming a very big problem in the United States, affecting 40% of people worldwide, and there is currently only one treatment approved by the Food and Drug Administration to treat MASH. A new class of drugs, based on FGF21 signaling, is showing good therapeutic benefits in clinical trials, but until now, the mechanism for how they work has been unclear,” said Potthoff, a professor of biochemistry and physiology at the University of Oklahoma College of Medicine and deputy director of OU Health Harold Hamm Diabetes Center.

The study’s results demonstrated that FGF21 was effective at causing signaling in the model species that changed the liver’s metabolism. In doing so, the liver’s fat was lowered and the fibrosis was reversed. The hormone also sent a separate signal directly to the liver, specifically to lower cholesterol.

“It’s a feedback loop where the hormone sends a signal to the brain, and the brain changes nerve activity to the liver to protect it,” Potthoff said. “The majority of the effect comes from the signal to the brain as opposed to signaling the liver directly, but together, the two signals are powerful in their ability to regulate the different types of lipids in the liver.”

Similar to the family of weight loss drugs known as GLP-1s (glucagon-like peptide 1), which help regulate blood sugar levels and appetite, FGF21 acts on the brain to regulate metabolism. In addition, both are hormones produced from peripheral tissues — GLP-1 from the intestine and FGF21 from the liver — and both work by sending a signal to the brain.

“It is interesting that this metabolic hormone/drug works primarily by signaling to the brain instead of to the liver directly, in this case,” he said. “FGF21 is quite powerful because it not only led to a reduction of fat, but it also mediated the reversal of fibrosis, which is the pathological part of the disease, and it did so while the mice were still eating a diet that would cause the disease. Now, we not only understand how the hormone works, but it may guide us in creating even more targeted therapies in the future.”