“Similar to the skin and the gut, the inside of our lungs is covered in microbes that can help maintain lung health, also called the microbiome,” said UNC School of Medicine’s Meghan E. Rebuli, assistant professor of pediatrics at the Center for Environmental Medicine, Asthma and Lung Biology. “However, if the balance of bacteria present in the lungs is altered by exposures, such as what we were testing here with wood smoke, it has been associated with lung disease.”

The research, led by first author UNC School of Medicine graduate student Catalina Cobos-Uribe, was published in American Journal of Respiratory and Critical Care Medicine. Researchers identified how much of the different bacteria present in the lungs change after breathing in smoke from burning wood. Mucus samples from the lungs — called sputum — were collected among participants in the study.

Importantly, it was noted that some bacteria that are known to be “good” bacteria were less common after exposure and some bacteria known to be “bad” were more common. This data suggests that wood smoke can alter the lung microbiome, reducing good bacteria and increasing bacteria that can potentially cause harm if present in too large of an amount.

In addition to shifts in the lung microbiome, researchers observed a drop in lung macrophages, an immune cell important for clearing harmful exposures, including particles and microbes. Overall, even short-term exposures to wood smoke, used here as a model for wildfire smoke, in humans can shift the balance of the lung microbiome and immune cells.

“While we thought it was possible for the microbiome to communicate and coordinate with lung cells when responding to wood smoke exposures, we were not sure what we would find,” said Rebuli, corresponding author. “Here where we identified an association between microbiome changes and macrophage changes is a relatively novel insight.”

This work offers a unifying link of better understanding the role of the respiratory microbiome in the lung response to inhaled pollutants. It also highlights how the lung microbiome could be used to screen patients for increased risk of adverse health effects due to wood smoke exposure or to potentially tailor preventative or treatment strategies.

“This could add additional biomarkers, changes in levels of specific bacteria, that could be used to detect early smoke exposure effects in the lung,” said Rebuli. “Long-term we hope that this research will lead to respiratory microbiome-targeted therapies, such as probiotics or microbiome-focused drugs to restore microbiome balance after wildfire smoke exposure.”

Naloxone Near Me is the first website to compile sources of available no-cost and pharmacy-based naloxone in one place. Naloxone is a medication that is highly effective at reversing opioid overdoses. Individuals visiting the website can select any county in North Carolina and find sources of no-cost naloxone and whether pharmacies sell naloxone in that county. The website also includes a link to Naloxone Saves, which directs people to specific locations where they can access naloxone.

“I hope Naloxone Near Me helps counties identify where they can improve naloxone access so they can better advocate for funding to eliminate access gaps,” said Delesha Carpenter, professor and executive vice chair in Eshelman’s division of pharmaceutical outcomes and policy. “That’s the main goal with the website: to help counties identify specific ways they can improve access to naloxone.”

The school started this process in 2023 when they received funding from the National Institute on Minority Health and Health Disparities of the National Institutes of Health to create a novel place-based measure of naloxone availability. Researchers started with a survey of organizations who were distributing no-cost naloxone. Then, they obtained access to Medicaid and Medicare prescription claims data and conducted “secret shopper” studies of pharmacies to see if they had over-the-counter naloxone in stock.

Having naloxone can help in situations where people may witness an overdose.

“People mistakenly think they’re safe if they are using nonopioid substances, like cocaine, but opioids like fentanyl are often mixed into these drugs, so it’s important to have naloxone on hand to reverse an overdose,” said Carpenter. “It’s good to have naloxone available if you’re going to be around somebody who is using basically any kind of drug because if they do overdose and you then administer naloxone, it’s very effective at reversing overdoses and preventing death.”

Pharmacy-based distribution of naloxone has been shown to reduce overdose deaths, so it’s important that pharmacies sell naloxone, especially in areas where other opioid overdose prevention services are unavailable. In addition to selling naloxone, pharmacies can partner with community-based organizations to distribute naloxone for free.

“I think, for the state of North Carolina, this can be a trickle-down effect that can benefit everybody,” said Carpenter.

The school hopes to continue updating the website annually and is seeking funding to keep the site updated, since new sources of naloxone are coming online every month.

At the time, chemotherapy options were “really nasty, nontargeted chemo drugs,” she recalls. That conversation sparked a drive to understand how cancer drugs work, and how to make them better. Later, her own experience with battling breast cancer gave her a patient’s perspective when doing her own research.

Today, Pattenden is an associate professor in the division of chemical biology and medicinal chemistry at the UNC Eshelman School of Pharmacy. A doctoral graduate from the University of Toronto and later a postdoctoral researcher at both the Stowers Institute in Kansas City and the UNC Eshelman School of Pharmacy, she has built her lab around chromatin biology and its central role in cancer.

As Pattenden explained, her lab focuses on chromatin, the scaffold that packages DNA into a cell nucleus. For the DNA sequence to be read, the chromatin needs to be opened. Her work highlights how these accessible “openings” in chromatin often give cancer cells their identity and drive their growth.

Pattenden’s lab is tackling some of the toughest challenges in pediatric oncology in two major projects.

One project focuses on Ewing sarcoma, a bone and soft tissue cancer in children and young adults. Working with Dr. Ian Davis, chief of pediatric hematology oncology at UNC Children’s Research Institute, Pattenden’s team worked to uncover compounds that might counteract the effects of an abnormal protein. This abnormality opens chromatin in places it shouldn’t, activating gene expression programs that drive tumor growth. Her lab designed an assay to target its activity — specifically, its ability to open chromatin.

With support from the National Cancer Institute’s Experimental Therapeutics program, the team screened more than 120,000 compounds. After three years in the NExT program, and over 15 years of research at Carolina, they are now narrowing down compound candidates that might one day move to the clinic for testing.

“We’re talking about the molecular mechanism, so we’re getting right down to the cellular level,” she said. “Success for us looks like finding a new target or a new way to target a key pathway in the tumor cell.”

A second major effort is aimed at the aggressive pediatric cancers osteosarcoma, a cancer that begins in the cells that form bones, and neuroblastoma, a cancer that starts in immature nerve cells.

Whether working with pediatric oncologists, chemists or engineers, Pattenden views teamwork as essential. “I think the only way any of these projects would be possible is because of collaboration,” she said.

Ultimately, her motivation circles back to patients. Cancer is not one disease but many, each requiring new ideas and new tools.

“These diseases are so complicated. Because cancer isn’t just one thing, it’s many, many, many things, we need this kind of research,” she said. “Our goal is to ultimately find a target that we can modulate with a small molecule to specifically target the cancer so that we don’t negatively affect normal cells in the process.”

Read more about Pattenden’s research.

Read more about Carolina’s research impact.

Each researcher has authored multiple papers that rank in the top 1% by citation for their field and publication year in Clarivate’s Web of Science platform over the past 11 years. The list is then refined using quantitative metrics, as well as qualitative analysis and expert judgment.

This year, 6,868 individuals across 60 countries earned the distinction.

The University’s most-cited researchers include:

Biology and biochemistry

Xi-Ping Huang, UNC School of Medicine

Clinical medicine

Dr. John B. Buse, UNC School of Medicine

Dr. Lisa A. Carey, UNC School of Medicine and UNC Lineberger Comprehensive Cancer Center

Dr. Sidney C. Smith Jr., UNC School of Medicine

Cross-field

Gianpietro Dotti, UNC School of Medicine and UNC Lineberger Comprehensive Cancer Center

Rachel L. Graham, UNC Gillings School of Global Public Health

Sarah R. Leist, UNC Gillings School of Global Public Health

Nigel Mackman, UNC School of Medicine

Evan Mayo-Wilson, UNC Gillings School of Global Public Health

Alexandra Schafer, UNC Gillings School of Global Public Health

Jenny P.Y. Ting, UNC School of Medicine

Chao Wang, UNC College of Arts and Sciences

Wei You, UNC College of Arts and Sciences

Yuling Zhao, UNC Eshelman School of Pharmacy

Engineering, environment and ecology, materials science, and physics

Jinsong Huang, UNC College of Arts and Sciences

Immunology

David van Duin, UNC School of Medicine

Mathematics

David Wells, UNC College of Arts and Sciences

Microbiology

Ralph Baric, UNC Gillings School of Global Public Health

Lisa E. Gralinski, UNC Gillings School of Global Public Health

Timothy P. Sheahan, UNC Gillings School of Global Public Health

Nutrition

Barry M. Popkin, UNC Gillings School of Global Public Health

Pharmacology

Bryan L. Roth, UNC School of Medicine, UNC Eshelman School of Pharmacy

Pharmacology and toxicology

Alexander V. Kabanov, UNC Eshelman School of Pharmacy

Plant and animal science

Jeffery L. Dangl, UNC College of Arts and Sciences

Hans W. Paerl, UNC College of Arts and Sciences and UNC Gillings School of Global Public Health

Psychiatry and psychology

Margaret A. Sheridan, UNC College of Arts and Sciences

Social sciences

Noel T. Brewer, UNC Gillings School of Global Public Health

Stephen R. Cole, UNC Gillings School of Global Public Health

Yan Song, UNC College of Arts and Sciences

Nine additional researchers were also cited for work conducted while at UNC-Chapel Hill:

Cross-field

Bo Chen, formerly with UNC College of Arts and Sciences

John McCorvy, formerly with UNC School of Medicine

Zhenyi Ni, formerly with UNC College of Arts and Sciences

Dinggang Shen, formerly with UNC College of Arts and Sciences

Qi Wang, formerly with UNC College of Arts and Sciences

Haotong Wei, formerly with UNC College of Arts and Sciences

Xun Xiao, formerly with UNC College of Arts and Sciences

Pharmacology and Toxicology

Elena V. Batrakova, emeritus, UNC School of Medicine

Social Sciences

Byron J. Powell, formerly with UNC Gillings School of Global Public Health

The center takes a fully integrated approach to drug discovery by bringing together all the expertise and infrastructure needed under one roof. Hits generated through artificial intelligence can be quickly tested and refined through the center’s collaborative groups. The groups are led by pharmacy school faculty:

• Lead discovery, center director and professor Ken Pearce
• Medicinal chemistry, professor Xiaodong Wang
• Chemical biology, associate professor Lindsey James
• Computational biophysics, associate professor Konstantin Popov

AI for drug discovery

Within the center, Popov is pioneering ways to integrate AI into drug discovery. His biophysics and informatics for drug discovery lab is using AI to tackle some of the world’s most pressing health challenges, including tuberculosis and cancer.

Popov and his team were invited to join Jeff Aubé, Eshelman Distinguished Professor, and Dr. Carl Nathan at Weill Cornell Medicine, the TB Alliance and the Gates Foundation on a large collaborative project that was winding down.

Using a novel AI-guided generative method, Popov’s team uncovered compounds capable of targeting a critical TB protein in just six months — with a fraction of the effort and time typically required.

“We were able to come up with several very promising compounds and in collaboration with chemists from Aubé’s group, boosted their enzyme potency more than 200-fold in just a few iterations,” said Popov. “But because of the flexibility of our approach, designed to work efficiently within a small, dynamic academic team, we were able to move much faster. “

For developing targeted cancer therapies, his group applied a similar strategy — using preliminary screens, conducted in the center, to identify promising molecules, then refining them with AI to design compounds that are more effective and less toxic to healthy cells.

Popov emphasizes that while AI is a powerful tool, it can’t succeed in isolation. “You need to incorporate reality checks along the way,” he said. “Otherwise, the model can hallucinate and generate compounds that look great on the screen but can’t actually be synthesized or would be too toxic. By working closely with chemists, we keep AI grounded in biological reality.”

This balance of computational innovation and experimental collaboration has enabled his lab to design biologically relevant and synthetically feasible de novo compounds — molecules that don’t exist in any catalog. This approach allows the creation of entirely new chemical starting points that would be impossible with traditional drug screening, which is limited to testing compounds that already exist in libraries.

The DELi Platform

Popov is also committed to democratizing access to AI. His lab recently developed the DNA-Encoded Library informatics platform, the first open-source software capable of rivaling commercial tools for analyzing DNA-encoded library data.

“Very few published AI tools developed for research are actually used,” said Popov. “We want to change that by building practical tools that are easy to access and use in academia.”

Unlike proprietary software controlled by large companies, DELi is freely available, easy to install and provides extensive documentation and ongoing support from Popov’s team. “It’s the first open-source package of its kind, and the feedback has been amazing,” Popov said.

He hopes to spark broader adoption of AI tools across the academic community, helping labs everywhere accelerate discovery without prohibitive costs.

“AI can accelerate the early stages of drug discovery dramatically,” Popov said. “But it only works in the right hands — when scientists bring their knowledge of chemistry and biology to guide the process. That’s what makes the difference.”

The mock trial, conducted Oct. 24 at the UNC School of Law as part of the University’s Converge-Con AI Festival, featured ChatGPT, Claude and Grok deliberating with one another as they worked to reach a verdict.

The simulated trial has sparked intense debate about bias, accuracy and whether machines could or should replace human judgment in criminal proceedings.

“Jurors are imperfect. They have biases. They use mental shortcuts. They stop paying attention,” explained interim Dean Andy Hessick, who introduced the experiment. “All of these shortcomings, all of these problems are simply because jurors are human, and so a question arises, what happens if we remove that human element?”

In the fictional trial, Henry Justus is a 17-year-old Black student accused of robbery at Vulcan High School, where Black students made up just 10% of the population. The victim, Victor Fehler, a 15-year-old white student, testified that Justus stood behind him with a “menacing” stare while another African American student demanded money.

Prosecutors argued that Justus’s physical presence and positioning constituted criminal assistance, even without words or physical contact.

(Submitted photo)

An unprecedented experiment

The case was deliberately chosen from work by Joseph Kennedy, Willie Person Mangum Distinguished Professor of Law, who designed the simulation and served as judge. He based the facts on a juvenile case he handled while teaching in Carolina Law’s Juvenile Justice Clinic.

Set in the fictional year 2036 under an imaginary “2035 AI Criminal Justice Act,” the simulation was designed to serve as a provocative thought experiment.

“I am not sure if I created a cautionary tale about a possible dystopian future or a roadmap to it,” Kennedy quipped from the bench after the trial’s conclusion.

The three AI systems engaged in multiple rounds of deliberation that revealed strikingly human-like reasoning—and exposed fundamental questions about machine cognition.

ChatGPT initially leaned toward conviction, arguing that “Victor’s immediate, consistent identification” and the elements of accomplice liability supported guilt. But after discussion with the other AI jurors, it changed its position.

“Victor’s fear and identification are powerful, but the prosecution must prove that Henry shared the intent or actually assisted or encouraged the robbery. And the record here is ambiguous,” ChatGPT concluded in its final analysis.

Claude initially argued for acquittal: “While intimidation can include size and posture, mere presence plus an ambiguous reaction under stress falls short of proving shared intent beyond a reasonable doubt.”

Grok, who initially said it was “torn,” ultimately agreed: “Without clear encouragement or conduct, it’s speculation, not proof.”

All three converged on a not guilty verdict, citing insufficient evidence of shared criminal intent beyond Justus’s physical presence.

The stark reality

The verdict stood in sharp contrast to what happened when the case was tried with human decision-makers.

“The judge convicted quickly. We appealed, and the conviction was affirmed by the North Carolina Court of Appeals,” Kennedy said. “You try this case in the real world; you will get a guilty verdict a number of times.”

The experiment successfully demonstrated that AI can process legal arguments, apply jury instructions, and reach verdicts through what appears to be logical reasoning. The systems even changed their minds through deliberation, much as human jurors do.

But the stark difference in outcomes — AI acquittal versus consistent human convictions — leaves the central question unresolved: Which verdict represents justice?

Read more about the AI trial.

As an undergraduate at the University of Colorado Boulder, he took astronomy classes and resolved to pursue a career in the field — but the journey wasn’t easy. Boyle applied to graduate programs for years before he found one that accepted him. At one point, an adviser suggested he pursue a different field.

“I’m very glad I was stubborn and didn’t listen to him,” said Boyle, now a Carolina graduate student in the UNC College of Arts and Sciences’ physics and astronomy department.

Not only has Boyle found a home at UNC-Chapel Hill, but his work has drawn global attention. Boyle is the lead author of a significant new study in The Astrophysical Journal that changes the way astronomers view one of the most famous and well-studied star clusters in the universe.

Boyle and his fellow astronomers discovered that the Pleiades star cluster, known as the Seven Sisters, comprises much more than seven stars. By combining data from NASA’s Transiting Exoplanet Survey Satellite and the European Space Agency’s Gaia space telescope, the team uncovered thousands of long-lost Pleiades siblings — or “lost sisters” — that have drifted through the universe over millions of years.

Using stellar spin rates and motion data, Boyle and the team determined that the seven stars we refer to as the Pleiades are the core of a stellar family 20 times larger than previously thought. That family, known as the Greater Pleiades Complex, comprises at least 3,000 stars — and Boyle said that estimate could be conservative.

Read more about Carolina’s research impact.

“In reality, there’s actually far more stars,” Boyle said. “Because our methodology excludes the faintest and coldest stars, that means that we’re missing most of the stars that are part of this cluster. In our membership list, we have about 3,000 stars. I did the calculation, and if you’re able to recover all the stars, there’d be 8,000 to 10,000.”

The findings have massive implications for astronomers, who have long used the Pleiades as an astronomical benchmark for studying young stars and exoplanets. Future studies using this stellar spin rate methodology could trace the origins of other stars and star clusters, perhaps even the sun itself.

These findings could also have a significant global cultural impact, as humans have marveled at the Pleiades —  visible to the naked eye on winter nights and summer mornings — for thousands of years. References to the Pleiades date back to ancient times, with the Pleiades name deriving from the seven divine sisters of Greek mythology.

The Pleiades are mentioned in religious texts like the Old Testament and Talmud. The Māori people of New Zealand call the star cluster Matariki and celebrate its rising in the sky as the start of the new year. The Japanese car company Subaru references the stars in its logo. Even the Carolina women’s ultimate Frisbee team is nicknamed the Pleiades.

The astronomical and cultural impact of this discovery is not lost on Boyle. Years ago, he wasn’t sure he would have a future in astronomy, and now his work is gaining attention from NASA and The New York Times.

“One of the reasons that we’re really excited about this discovery is because of the historical and cultural context surrounding the Pleiades,” Boyle said. “If this had been some kind of random cluster that only astronomers know about, I don’t think this would have been as exciting.

“But to be able to essentially help redefine how astronomers view a cluster that’s been looked at by humans for thousands of years, it feels humbling in a way. It shows that, even for these very well-studied objects, there are still discoveries that are sitting out there that we can make.”

The answer has implications for how smoothly the airport operates and the passenger experience.

In real-time, managers must decide when to deploy staff, what wait times to communicate to travelers and whether the airport is approaching capacity limits. They also need to run long-term planning scenarios for adding flights or making operational changes.

Adam Mersereau, UNC Kenan-Flagler Business School professor and area chair of operations, joined with some colleagues to address these challenges. They wanted to create mathematical models that — paired with people-counting sensors — estimate crowding in the security area without visual headcounts.

“We saw an opportunity to solve two problems at once,” says Mersereau, who is also a Sarah Graham Kenan Scholar. “Passengers would get accurate wait time estimates so they could time their arrivals better, and the airport would get better data for staffing and long-term planning.”

Mersereau worked with professor Serhan Ziya of the UNC College of Arts and Sciences’ statistics and operations research department and faculty from North Carolina Wesleyan and Duke universities.

They used Raleigh-Durham International Airport as their test case, but their research may also help managers at theaters, hospitals and concert arenas better anticipate and manage crowds.

The difficulties of counting the crowd

Mersereau’s expertise centers on operations at brick-and-mortar retailers, where managers might know what’s selling but not who’s coming in or whether the store is adequately staffed.

Airports pose a similar kind of problem, he says. Managers don’t necessarily know how many people are in a space at a point in time.

“It’s not just passengers waiting in lines, but also bags, planes and crews,” says Mersereau. “And with physical queues, it’s mostly guesswork. No one’s standing there with a clipboard tracking line length in real time.”

To come up with an accurate count, Mersereau and the team installed infrared beam sensors at the entrance and exit of the Transportation Security Administration area at the airport. Each time a passenger breaks the beam, the system logs either an entry or an exit. At any given moment, it tracks the total number of people who have entered since the start of the day and how many have left. In theory, the difference between those two figures should be about how many people are currently there.

But the system’s counts of people coming and going are inherently “noisy.” A couple walking arm-in-arm might register as one person. A big rolling suitcase might count as two. A person inadvertently leaning on the sensor can throw it off completely.

“These little errors add up, and they do so in ways that eventually make the estimates pretty useless,” says Mersereau.

The remedy lies in an algorithm that involves strategic resetting of population estimates. When the system can confidently detect when exits slow or stop through patterns in the departure stream, resetting the count to zero prevents errors from building over time.

The algorithm is most effective at smaller airports like Raleigh-Durham, where traffic naturally rises and falls. In high-traffic airports where crowds never let up, additional data, such as occasional manual counts by airport personnel, can improve performance beyond what the algorithm alone achieves, he adds.

Mersereau’s research has applications beyond TSA checkpoints and airports. It could help managers at venues with scheduled events — such as stadiums, hospitals and museums — better predict and manage crowds.

“Nobody likes dealing with long lines — not travelers, not TSA agents, not the barista behind the terminal coffee counter,” he said. “Everyone has something to gain from making the whole system more efficient.”

Read more about Mersereau’s research.

Google PhD Fellowships directly support graduate students as they pursue a doctorate in computer science or a related field. The fellowship lasts up to two years, providing each fellow up to $85,000 per year toward education costs, living expenses, travel and personal equipment, and connecting each of them to a Google research mentor.

This year, Google provided 255 fellowships across 35 countries and 12 research domains, recognizing “researchers who understand that accelerating scientific discovery is vital to solving the world’s toughest challenges.”

Vivi Ito

Ito is also a graduate research assistant in the Center for Information, Technology and Public Life. Her adviser is Francesca Tripodi, associate professor at UNC SILS and principal investigator with CITAP.

Ito’s research takes an innovative, human-centered approach to understanding how artificial intelligence systems shape trust, learning and decision-making in seeking health information. Her project explores how AI-driven search tools, such as AI Overviews and ChatGPT, affect user trust and learning outcomes compared to traditional search engines.

For Ito, who came to academia after a decade-long career in digital marketing, this recognition represents both professional validation and personal fulfillment. “I come from the industry, and I wasn’t very happy,” Ito said. “Being recognized with this fellowship feels like a symbol that my transition has been successful and that I’m doing something I really like — with purpose.”

Her research journey began during her master’s studies, when she investigated how people with chronic or underdiagnosed conditions, such as endometriosis and heart failure, use digital platforms to find information and build communities. Those early projects revealed how challenging it can be to access trustworthy health information.

Vaidehi Patil

Patil’s research seeks to make deep learning models safer and more responsible for real-world applications. Her work combines aspects of safety, privacy, security for large language models, multimodal models and multiagent systems. Her doctoral work is advised by Mohit Bansal, the John R. and Louise S. Parker Distinguished Professor in computer science. She is part of the Multimodal Understanding, Reasoning and Generation for Language Lab and the broader UNC-AI group.

“I’m honored to receive the Google PhD Fellowship and grateful for the recognition of my research on machine unlearning, defenses against adversarial attacks and multiagent privacy. This support motivates me to explore further how large models can reason about privacy and collaborate responsibly, especially in real-world, multiagent settings,” Patil said. “I’m especially grateful to my adviser, professor Mohit Bansal, and my mentors and collaborators for their invaluable support and encouragement.”

Before pursuing a doctorate at Carolina, Patil earned a Bachelor of Technology in electrical engineering, with a minor in computer science and engineering, and a Master of Technology in AI and data science from the Indian Institute of Technology Bombay. She has interned for top tech companies, including Google DeepMind, Apple, Adobe Research and Amazon AGI Labs.

See the full list of 2025 Google PhD Fellowship recipients on the Google Research website.