On the first floor of the Bronx High School of Science, in the hallway located outside of the main office, stretching from the Physics Wing to the Small Gym, numerous wooden and metallic plaques line the walls. For the past 83 years, they have commemorated the Bronx Science students who have been named Semi-Finalists and Finalists in the nation’s most prestigious and respected science research competition.
This year, twelve seniors will join the storied history of Bronx Science and the Regeneron Science Talent Search, receiving recognition for their outstanding research projects. These Regeneron Scholars are in the top 300 of 2,471 students worldwide who submitted to the competition for 2025. The Society for Science, which runs the competition, reports that this was the largest applicant pool since 1967.
This year, Bronx Science was the high school with the most Scholars (the contest’s designation for Semi-Finalists), out of almost 800 secondary institutions from 48 states and 17 other countries. While the Regeneron Science Talent Search is exclusively open to high school seniors in the United States or living abroad with U.S. citizenship, the Society for Science also runs the International Science and Engineering Fair, in which high school students of all grades and from around the world are eligible to submit their entries.
Mr. Richard Lee, a Biology Research teacher at Bronx Science with three Scholars in his class this year, has been teaching research and mentoring research students at Bronx Science for the past 27 years. “I knew coming in with this year’s research students, and through working with them this year, that they were going to be some who were recognized,” he said. With twelve Bronx Science Scholars for 2025, this year’s number of Semi-Finalists is the largest number for Bronx Science since 2021, when there were 14.
The Regeneron Science Talent Search (from 2017 to today), formerly known as the Westinghouse Science Talent Search (from 1942 with its inception to 1998), and the Intel Science Talent Search (from 1998-2016) recognizes high school seniors who have completed novel, impressive research projects in science and mathematics. In recent years, the application review process has become more holistic, taking into account the student’s high school transcript, extracurriculars, and essays.
This year, the projects of the Bronx Science Semi-Finalists range from an examination of online exposure to climate change messaging to approaches for predicting the crystal structure of biomolecules. Each of the awarded Scholars, or Semi-Finalists, will receive $2,000 for themselves and an additional $2,000 to fund STEM-related activities at Bronx Science. On January 23rd, 2025, the top 40 of the 300 in the nation will be announced as Finalists, each receiving a $25,000 cash prize. In March 2025, these finalists will travel to Washington D.C. to present their project to thousands of visitors, government officials, and top researchers. A judging panel then selects the top ten entries nationwide, with the first place winner receiving $250,000. In total, the Society for Science provides recognized students and schools $3.1 million in awards.
All of the selected Semi-Finalists have participated in the Bronx Science research program, which offers specialized tracks in Biology, Physical Science, Social Science, and Mathematics over a three year course of study. Of the 12 awardees, seven projects were in Biology, three in Mathematics, and two in Social Science.
Bronx Science aims to make research accessible to all, regardless of their background or previous experience. “Sometimes, the best student researchers are the ones who do not see themselves as ‘science people,’” said Dr. Tracy LaGrassa, one of the Biology research teachers and mentors. “The most innovative ideas come from students who are curious and unafraid to dive in the deep end. If you’ve experienced failure in your life, and managed to learn from it, then you’ve got the disposition of a researcher!”
Below are spotlights on the twelve 2025 Scholars, summaries of their projects, as well as their reflections on the research experience.
Vimala Albert ’25- Would a Rose by Any Other Name Really Smell Just as Sweet?: The Effect of First and Last Name Fluencies on Candidate Perception
Research field: Cognitive Psychology
Research teacher: Mr. Joshua Fialkow
Research mentor: Dr. Monica Biernat, University of Kansas
“My entire life, people have had difficulty pronouncing my first name. These experiences encouraged me to explore the effects of people’s names on their lives.
Every time your brain processes information, it associates a difficulty level with the act of processing that stimuli. The more difficult information is to process, the more negatively you will perceive the stimuli. This phenomenon applies to all stimuli, including names. The more difficult it is to pronounce (less phonologically fluent) a name, the more you will view the name-holder negatively, whether the name-holder is a person or a pharmaceutical drug. Most research on this “name-pronunciation effect” focuses on one name component. However, people generally have at least two — a first name and a last name.
I decided to use an election scenario to explore the issue largely in part due to the recent presidential election, where the candidates had different phonological fluencies. I found that individuals with a combination of fluencies (have a fluent first name and disfluent last name or a disfluent first name and fluent last name) are the best perceived and are more likely to be voted for in an election.
I’m confident that my research will help us, as a society, better understand how we perceive and judge others in our daily lives. I hope that my research encourages people to be more aware of their own internal biases as they interact with others. I wholeheartedly believe that if people are able to become more aware of their biases, they will interact with others in a more open-minded manner that will lead to more collaborative and communicative relationships.”
Connie Chen ’25 – A Novel Mathematical Framework for Crystal Structure Prediction
Research field: Crystallography
Research teacher: Mr. Richard Lee
Research mentor: Dr. Nikos Galanakis, NYU
“I conducted my research remotely since my work was all just computational. It was difficult to acclimate to this research, since I had to learn a completely new field of study.
Crystal structure prediction (CSP) has been an important endeavor, especially in the pharmaceutical industry. However, current methods rely on physical interactions between atoms of analyzing and predicting these structures, which can be unreliable in some cases. Instead, I approached CSP with a mathematical lens to hopefully increase its efficiency and accuracy. Crystals form the foundation of many drugs, but they exhibit a behavior called polymorphism where structures can be subject to change either in its production or even dissolution in the body. My project addresses this issue by proposing a more concrete way of finding the most stable structures, so the crystals don’t just suddenly change form and can perform their designated purpose.
I chose this particular problem because medicine is consumed by everyone, so this is an issue that can affect anyone. I hope this discovery can improve the pharmaceutical industry.”
Myles Coven ’25 – Novel Regulation of Low-Density Lipoprotein Receptor (LDLR) by UBOX5
Research field: Cellular and Molecular Biology
Research teacher: Mr. Richard Lee
Research mentor: Dr. Leigh Goedeke, The Icahn School of Medicine at Mount Sinai
“The idea of conducting research at Mount Sinai was very daunting to me. However, the amazing team made the environment extremely welcoming. I was concerned that a high school student would not have much in common with those in my lab much older than me; however, we quickly found common ground and became very close.
Atherosclerosis is a disease in which a type of cholesterol, known as low-density lipoprotein (LDL) cholesterol, accumulates in the arteries. Influenced by genetic and environmental factors, Atherosclerosis is the leading cause of morbidity and mortality worldwide, and has been shown to lead to heart attack and stroke.
Regulating LDL-cholesterol has been of interest to researchers for decades in order to develop therapeutics to decrease disease risk. One primary target has been hepatic LDL-receptors (LDLR), which sense and uptake LDL-cholesterol.
Recently, a gene called UBOX5 has been identified as a top regulator of LDLR activity. When the expression of UBOX5 was silenced, there was a significant increase in expression of LDLR. This was later shown to be dependent on a regulator of LDLR known as SREBP2. When SREBP2 was inhibited, even without UBOX5, expected LDLR increases were not seen. In my research, I found that the knockdown of the gene that I was focusing on, UBOX5, leads to significant increases in LDLR activity and is only possible through SREBP2.
UBOX5 inhibition could be a new way to treat high cholesterol in addition to statin therapy, being a new therapeutic mechanism for combatting atherosclerosis.
Being a Regeneron STS scholar means so much more to me than just the name. It represents being able to share my passion for research with others and makes me feel incredibly happy that I am able to achieve this simply doing what I love. It has reinvigorated my drive for research, and I am more motivated than ever to get back in the lab and continue to deepen our understanding of biology. I love to take photographs in my free time, going on long walks and photographing the city. I also love playing piano, frequently teaching myself pieces by my favorite composer Chopin.”
Benjamin Gao ’25 – Connecting the Tropics to Ice Ages: Exploring Geochemical Indicators in Sea Foraminifera To Reconstruct Paleoclimate Sea Surface Temperatures in the Western Pacific Warm Pool
Research field: Climate Science and Geochemistry
Research teacher: Dr. Vladimir Shapovalov
Research mentor: Dr. Braddock Linsley, Columbia University
“Through studying ancient ocean temperatures, my project focuses on understanding how Earth’s climate has changed over hundreds of thousands of years. Specifically, I examined a region in the Western Pacific called the Warm Pool, which holds the largest area of warm water on Earth and plays a key role in distributing heat and regulating global climate. Climate shifts, such as the transition between ice ages and warmer periods, are partly due to Earth’s natural cycles.
Studying these cycles help us to understand how modern human-driven changes may impact the planet. To explore these past climate shifts, I analyzed tiny foraminifera shells found in sediment cores taken from the warm pool. The chemistry of these shells, especially the oxygen isotope δ¹⁸O, a well known temperature tracer, provides clues about sea surface temperatures and ice volume changes over time. By measuring variations in δ¹⁸O over time, I illustrated temperature shifts in the Warm Pool and connected these patterns to global climate changes.
Understanding the significance of past climate shifts is essential for addressing the challenges posed by modern-day climate change. My project provides a higher resolution of the Earth’s paleoclimate data, allowing us to tackle climate change with the necessary data of how Earth’s climate changed in the past.
My experience in discourse through activities like debate and campaigning allowed me to understand the importance of climate research and its implications on handling today’s most pressing environmental challenges. Research has taught me how to think critically and has given me a deeper appreciation for the role that science plays in addressing societal challenges. Outside of research, I love discourse and politics. I’ve volunteered for political campaigns and worked for the New York Democrats. I am also a member of the Bronx Science Speech and Debate Team and President of the school’s Political Discourse Club.”
Liza Greenberg ’25 – Making Order With Disorder: Specific Structural Roles of Conformationally and Chemically Heterogeneous Glycopeptides Within Mucin Proteins
Research field: Structural Biology
Research teacher: Mr. Richard Lee
Research mentor: Dr. Deborah Fass, The Weizmann Institute of Science
“The process of conducting research was the most adventurous experience of my life. I traveled to the Weizmann Institute of Science in Rehovot, Israel, to work with my amazing research mentor Dr. Fass, who despite being a stranger, graciously hosted me in her home. While I learned to navigate the lab, which included lots of spills, puzzling over machinery, and spending hours trying to get molecular simulations to run, I also found myself trying to navigate a new country and culture on my own. The whole experience, including the host of setbacks and complications, have boosted my confidence immensely and taught me how to work responsibly and professionally. The experience was above all, made possible with the guidance of Dr. Fass, Mr. Lee, and all of the friends and family who have supported me.
My project is on the topic of mucus, which may seem gross, but in actuality, it is fascinating. Without mucus, every time someone ate, drank, or even breathed, they would die. Mucus is a barrier to the external world that plays crucial roles in defending against disease and protecting internal epithelial surfaces. My project investigated disordered segments of mucin proteins that are the primary component of mucus. These regions were thought not to play significant roles in mucin structure and assembly, but by comparing amino acids across species, visualizing mucins with electron microscopy, and performing molecular dynamics simulations, my research suggests that the amino acid sequence and length of these disordered segments may play a significant role in allowing mucus to properly assemble.”
Jayden Lin ’25 – Novel Interpretable Model for Urban Heat Island Forecasting and Addressing Climate Racism in Big Cities
Research field: Computational Civil Engineering
Research teacher: Dr. Vladimir Shapovalov
“With an estimated two-thirds of the world’s population living in cities by 2030, many people will be extremely susceptible to what are known as Urban Heat Island (UHIs). The UHI effect occurs in urban areas, which can be upwards of 7 degrees Fahrenheit warmer than surrounding rural areas. Energy demands in the United States are growing, and urban heating makes it worse with demand in cities increasing around 3% per Celsius. Cities also contribute 60% of greenhouse gas emissions with Air Conditioning units alone almost using 10% of global electricity and contributing to 3% of emissions. There is also evidence that indicates that UHIs affect minority populations disproportionately, as correlations were found between urban heating and racist housing policies.
To understand what specific features cause the UHI effect, I made an interpretable model to predict UHI event intensity. The model was optimized and tested, resulting in over 95% accuracy in predicting UHI event intensity. It shows promise for city planners in determining the best tools in order to reduce UHI event effects.
There are significant health risks to the population due to UHIs. Heat stroke, heat exhaustion, and heart attacks have all been correlated to UHIs, due to periods of heatwaves harming health and worker productivity. I also focused on the social aspects of urban science, as I’ve always wanted to combat urban inequities that exist in today’s society.
Urban science is something that I am extremely passionate about and hope to investigate further as I continue my research adventure in college. I would not have been able to accomplish this if it weren’t for the support of my friends. Thank you to Jonathan Iskhakov ’25, Kenneth Wu ’25, and any of my other peers that have been with me through everything. It means a lot.”
Sidney Lin ’25 – Machine Learning for Identification of Fast Progressors of Infarct Growth in Early Window Acute Ischemic Strokes Without Perfusion Imaging
Research field: Stroke, Data Science, and Epidemiology
Research teacher: Dr. Tracy LaGrassa
Research mentor: Dr. Kory Byrns, New York Presbyterian Queens
“Stroke is a leading cause of disability and mortality. Most strokes are ischemic, caused by a clot in a blood vessel located in the brain. When a patient arrives to the ER, data is gathered such as age, the time since symptom onset, the severity of the stroke, patient’s health history, laboratory tests, and imaging tests. In the early time window from stroke onset (less than 6 hours), this set of data is enough to make a decision about whether or not a patient should have a thrombectomy, a procedure needed to remove the clot and restore blood flow. However, it is controversial whether or not an additional specialized brain scan called CT perfusion could also add value. CT perfusion shows how much of the brain is already lost and how fast the stroke is progressing.
There is a dangerous subtype of stroke called “fast progressor,” which requires more caution before performing thrombectomy, because it has decreased rates of favorable outcomes. Fast progressors are usually diagnosed using CT perfusion. But, CT perfusion is an imperfect exam, since it is extremely sensitive to patient motion that could make the scan inaccurate, takes additional time to perform and interpret, requires more radiation exposure for the patient, and increases the costs to our healthcare system.
So, my project aimed to determine if machine learning algorithms can be trained in order to accurately identify fast progressors using only the routinely collected data, without the need for CTP. If possible, this could help streamline patient triage and thrombectomy selection, save time, costs, and radiation exposure, and serve as a backup for CTP. Machine learning could also act as an alternative for under-resourced hospitals that aren’t capable of performing CTP. My research was conducted using an anonymized database from my regional hospital.
I was inspired after attending a symposium on AI and stroke care at the 2024 International Stroke Conference. I wondered afterwards if the future of medical diagnosis could be machine-based and began brainstorming ideas regarding how machine learning can potentially help stroke diagnosis in the ER. I discovered that machine learning can identify fast progressors of infarct growth in early window anterior circulation ischemic strokes using routine baseline demographic, clinical, laboratory, and imaging data, without input from CT perfusion. The best performing model, XGBoost, did so with 89% accuracy.
My favorite parts were giving presentations of my work over the summer to my research class. It has been an amazing and surprising experience meeting so many young people who are also interested in machine learning and want to join in on the AI revolution that is beginning to make its way into every sphere of our lives. I was shocked and incredibly honored [when hearing of my 2025 Regeneron Scholars award]. It is a dream come true, since STS has such a long and rich history of excellence.”
Mary Loukaitis ’25 – Quantifying MAB-3-Controlled Robustness in Caenorhabditis elegans Male Tail Morphogenesis
Research field: Developmental Genetics
Research teacher: Dr. Tracy LaGrassa
Research mentor: Dr. Karin Kiontke and Dr. David H. A. Fitch, New York University Center for Developmental Genetics
“I wanted to gain experience in scientific research and experience it outside the classroom. Most importantly, I hoped to gain an understanding of biological systems as interconnected, coordinated cells or molecules rather than a series of disconnected factors. Laboratory experience provided an opportunity to analyze organisms through this lens, rather than through simplistic textbook diagrams.
I was inspired to enter the field of developmental biology by observing mitosis in a Caenorhabditis elegans embryo that I isolated at The Stanley Manne ’52 Research Institute™. The elegant widening and division of embryonic cells under 400x total magnification compelled me to understand the coordinated mechanisms behind biological function and life. With this C. elegans male tail morphogenesis study, I was grateful to perceive development as a sum of interacting parts.
I investigated biological robustness in C.elegans male tail morphogenesis. Biological robustness is the ability of organisms to maintain function under stress. To investigate robustness, I tested male tail development in the model organism, the microscopic roundworm Caenorhabditis elegans. The redundant MAB-3 transcription factor is suspected to govern robustness in C. elegans tail development, so I tested the effects of stress on null mab-3 mutants. These mutants normally have mild abnormal phenotypes, such as the formation of a pointy tail, reduced ray development, and short spicules. Under environmental and developmental stress conditions, however, mab-3 mutant phenotypes increase in severity. To explain this phenomenon, I created a network of protein-protein interactions to model the interactions and robustness of mab-3 and stress response pathways. Here, I show that MAB-3 may play a role in the robustness of tail development through redundant architectures and (tentative) links to robust stress response pathways. In the context of cancer, these findings can predict the effectiveness of eliminating redundant factors to hinder tumor stress resistance.”
Joshua Manasse Cammerman ’25 – Phagocytosis of Reticulocytes by Splenic B Cells in Malaria-Infected Mice
Research Field: Cell Biology and Immunology
Research Teacher: Dr. Emily Schmidt
Research Mentor: Dr. Eldad Hod, Columbia Presbyterian Irving Medical Center
“My research investigates how B cells (a type of immune cell) help control malaria infections through the removal of reticulocytes-young red blood cells-which certain malaria parasites prefer to infect. Malaria, a disease carried by mosquitoes, impacts millions worldwide and often causes fatal illnesses in children, resulting in hundreds of thousands of deaths each year. The Plasmodium parasite that causes the disease uses red blood cells to reproduce, leading to cyclical fevers and severe illness.
My study centered on the interactions between B cells in the spleen of malaria-infected mice with reticulocytes. Using specialized imaging and analysis tools that I created, I observed the consumption of infected reticulocytes by these B cells, suggesting B cells may have a function outside of antibody production. This behavior might be an evolutionary response to malaria: by clearing reticulocytes, the body may limit the parasite’s ability to spread.
For me, it was easy to adjust to my first year in the lab, because there were multiple other students there. It is an unusual experience, somewhere between camp and school, where you set your own expectations. The most surprising and exciting experience was getting a tour of the Blood Bank, where I got to see how the work done in my lab applied to the lives of the millions of patients that require blood transfusions. I have learned so much about immunology and blood transfusions, which is fun whenever I get blood drawn! Also, I recognize the incredible difficulty and work that research is, often taking years without success or breakthroughs. It is an incredibly long and arduous process, often with little to no reward, so when I found out that I was selected as a 2025 Regeneron Scholar, I was extremely excited and humbled.”
Christopher Procaccino ’25 – Breaking Barriers: Verifying Sterility of Germ-Free Humanized Mice
Research field: Immunology and Biotechnology
Research teacher: Dr. Khaled Mahmoud
Research mentor: Dr. Jian Guan, NYU Langone Department of Cell Biology
“As someone who has always craved independence, in the classroom or in my personal life, the research program offered me the perfect amount of independence, demanding of me to take the initiative to find a research lab while also having extensive support from adults in the program.
It was difficult to acclimate in my circumstances, because I had to switch labs in the middle of my junior year. My mentor unexpectedly left my first lab around December during my junior year, meaning that I had to restart the cold e-mail process and quickly find another lab in which to conduct research.
My project addresses the all-too-common discrepancies between the mouse and human immune systems that often cause drugs that prove successful in animal trials to fail in clinical trials. By making mouse models more closely resemble the human immune system, fewer resources will be wasted on drugs that are incompatible or harmful to the human immune system. In addition, humanized mouse strains that aren’t germ-free still present the problem of bacteria, fungi, and other microorganisms that could complicate results, acting as omnipresent confounding variables.
My proudest moment was opening the qPCR graph to see that my experiment had worked – the mice were indeed germ-free. It was a small discovery, but in the cold, unmerciful air of the lab, it felt enormous. I realized that years of developing my lab skills and poring over immunology textbooks had finally come to fruition.
Another one of my favorite moments was sitting in on three-hour-long lab meetings that never truly felt like three hours. Discussing discoveries from the lab’s dozens of members, I was enthralled by how enormous and undiscovered the field of immunology was, and I was persuaded to learn more every single day.”
Michael Raziel ’25 – Decoding Denial: Using a Granger Causality Model To Examine the Impact of Climate Consensus Messaging on Climate Change Discourse in Online Platforms
Research field: Behavioral Science
Research teacher: Dr. Scott Saviano
“As a member of a nonprofit, We Care Act NYC, dedicated to stopping the proliferation of environmentally harmful electronic waste, I was curious about how social media could be utilized to encourage more sustainable behaviors in those exposed to our content.
Although I don’t attend an actual lab, consistently communicating with field professionals who are working on the most innovative research in their respective fields is an eye-opening experience. I cherish being surrounded by intelligent and motivated individuals who give me an idea of how far I can truly take my project.
My research aims to find a low-cost, quantitatively rigorous, and real-time method to analyze important online sentiment trends, rather than relying on current survey-based methodologies, in order to increase scalability. I found a low-cost method that validates the theory that exposure to messages related to scientific consensus on climate change in online platforms can influence users’ sentiment towards climate change. Using a dataset containing 4.6 million online comments about climate change and a Python algorithm that I built with my research teacher, Dr. Savaiano, I created a Granger causality test to analyze correlations with a time series lag, enabling us to examine causal relationships.
My larger hope is that these findings can be extrapolated to provide comprehensive models and strategies that organizations and campaigns can use to frame their anti-warming messages in the most effective ways in online spaces. This will ultimately impact people’s attitudes toward climate change and help to curb climate change denial.
My best memories were definitely working with Dr. Savaiano during his SGI periods (office hours) to figure out how to tweak the methodology multiple times, in order to ensure that it was as effective as possible.”
Miranda Zhao ’25 – Extreme Weather Events Predicted by Novel Active Machine Learning Model With Deep Neural Operator
Research field: Computational Climate Modeling
Research teacher: Dr. Vladimir Shapovalov
“Extreme weather events have always been a part of my life, whether it is by walking home from school in the orange air of New York City from Canadian wildfire smoke in June 2023 or calling relatives in Florida during the recent hurricanes Helene and Milton. I saw firsthand that the subsequent devastation was so strong because (amongst other factors) prediction of these events was not very good. Specifically, the temporal accuracy (i.e., when the storm is going to hit, and how long it is going to be there) wasn’t accurate enough for people to adequately prepare enough to leave, and for larger organizations to prepare the local infrastructure. Moreover, I wanted to dig into how regional events could affect global weather patterns (such as the Canadian wildfires spreading to New York City).
The project that I submitted modeled extreme weather events (both long-term and short-term, such as droughts and hurricanes) through deep neural operators, which are able to create input-output pairs between initial conditions without a lot of preexisting input-output pairs. The deep neural operators in my model were able to predict extreme weather events on the order of a week. I used my personal computer for the programming and for running my program (it was very resource-intensive).
My novel machine-learning model provides proof-of-concept that the process of using deep neural operators and probabilistic functions sidesteps previous issues relating to machine learning regarding extreme weather prediction in medium-term timescales, such as computational expense, lack of sufficiently accurate data, potential overfitting from this lack of data, and the inability to take unknown effects of climate change into account. This research could potentially help with more accurate predictions of an extreme weather event’s timespan and severity. My hope is that these can be used to help people escape these events, as they only grow stronger and more frequent due to climate change.”
“Being a Regeneron Science Talent Search Scholar means so much more to me than just the name. It represents being able to share my passion for research with others and makes me feel incredibly happy that I am able to achieve this simply doing what I love,” said Myles Coven, a 2025 Regeneron Semi-Finalist whose research focused on combatting atherosclerosis.