ARTIFICIAL MADE REAL

Villanova Engineers harness AI to transform research into solutions that shape the future

By Andrew Faught

Artificial intelligence is reimagining the world as we know it. The technology curates online content, pilots self-driving cars, manages supply chains and helps to predict heart attacks.

None of it would happen without an engineer.

At the College of Engineering, faculty and students are helping to reshape industries, economies and everyday life. They’re building systems that could lead to exciting new drug treatments, protect communities from extreme weather and find new clean energy sources.

The next wave of innovation will be the result of thousands of engineering decisions involving hardware, systems design, optimization and reliability—decisions that will transform mathematical ideas into real-world tools.

These Villanova faculty members are showing us the way.

COMBINING DATA, SCIENCE AND MACHINE LEARNING

In Villanova’s Multiscale Modeling of Materials and Machine Learning Laboratory (M4L Lab), students use AI to create and validate computational models that address diverse challenges, such as identifying next-generation materials for fusion energy applications, predicting stress fractures in athletes and designing advanced dental implants.

As principal investigator of the lab, David Cereceda, PhD, associate professor of Mechanical Engineering, reminds his charges not to lose sight of the bigger picture.

“I combine both the physics-based models developed over the past couple of centuries with modern data-driven approaches,” he says. “I tell my students that we don’t want to treat AI and machine learning as a black box. We want to build on the knowledge we already have and use AI tools to expand our understanding and accelerate discovery.”

It’s an approach that has served Dr. Cereceda well in his own work on materials engineering for nuclear fusion, a potential clean energy source that scientists hope could power cities with minimal carbon emissions. In 2022, he was Villanova’s first-ever recipient of the US Department of Energy’s prestigious Early Career Award, for his project “Unraveling transmutation effects in tungsten-based plasma facing materials: a computational approach that integrates nuclear transmutation, first-principles, calculations and Machine Learning.”

Fusion reactors must operate under extreme conditions: temperatures reaching thousands of degrees, intense magnetic fields, and bombardment from high-energy neutrons. Such an environment can degrade materials quickly, making durability a persistent challenge in fusion technology. By combining physics-based simulations with machine learning models, Dr. Cereceda can predict which materials are most likely to withstand such extreme environments.

His AI work also has a deeply human dimension. Inspired in part by personal experiences that strengthened his commitment to translational research with real societal impact, he collaborates with clinicians to develop decision support systems that analyze patient data and reveal patterns that might otherwise go unnoticed.

One project uses machine learning to predict spontaneous preterm birth, a major cause of infant health complications worldwide. In another, he works with the Gait Lab at Nemours Children’s Hospital to develop a clinically informed machine learning framework that enhances decision support tools for ambulatory youth with cerebral palsy. The team’s approach integrates patient-specific data and clinical expertise to predict gait kinematics, identify key predictors and characterize treatment responsiveness across different interventions and age groups, from early childhood through early adulthood.

“These systems are tools,” Dr. Cereceda says. “They help clinicians identify risks earlier, but the final decisions always belong to the medical professionals.”

TRACKING THE SPREAD OF DISEASE AND DISRUPTIONS

When COVID-19 emerged in China in December 2019, it took just nine months for the virus to spread to every corner of the globe.

Chenfeng Xiong, PhD, assistant professor of Civil and Environmental Engineering, is using AI to study transportation big data, human mobility research and travel behavior modeling to better prepare communities and governments for future pandemics and disasters.

Dr. Xiong is developing models to re-create the daily movements of nearly 20 million people in Lagos State, Nigeria. He’s attempting to predict how diseases, disasters and disruptions spread through human populations, in an attempt to stop them and save lives in a timely manner. Dr. Xiong is also building models around American air traffic.

“We’re envisioning an increase in situational awareness, especially for those areas that are not as data-rich as the United States,” he says. “They can then provide a more timely response, instead of sitting in the dark and waiting for the rest of the world to take action.”

Beyond air travel data, Dr. Xiong is using railway station and bus hub counts, road traffic measurements and anonymized smartphone location data to build his models.

At the center of the system is AI-driven deep learning, including large neural networks—a machine learning model inspired by the human brain. The technology uses algorithms to predict a person’s likely daily itinerary, including when they leave from home and return, and by what mode they travel.

“Before AI, we could never model all these factors in one step,” Dr. Xiong says. “Now, deep learning allows us to handle extremely complex patterns.”

By combining AI with real-time information sources, the model could detect an emerging disease threat, simulate its spread and identify hotspots before an outbreak, Dr. Xiong adds.

Authorities could then take targeted actions, such as local lockdowns, healthcare deployment or vaccination drives, focused on areas of greatest need.

“I think in five years, this simulation model will work within hours or minutes to make predictions,” Dr. Xiong says. “We’re trying to promote accurate information with AI’s growing power.”

PREDICTING—AND PREVENTING—URBAN FLOODING

As extreme weather becomes more frequent and intense, urban flooding is becoming a growing threat. But AI algorithms—problem-solving rules fed to a computer—could translate into decreased stormwater damage, increased infrastructure longevity and improved equity in how government agencies protect their communities.

In a multidisciplinary partnership, Virginia Smith, PhD, associate professor of Civil and Environmental Engineering, is working closely with Xun Jiao, PhD, associate professor of Electrical and Computer Engineering, to investigate how green stormwater infrastructure (GSI)—from porous paving to sidewalk planters and “rain gardens,” which filter pollutants and help to replenish groundwater—can be managed to minimize risks before destructive weather hits.

“We’ve been using AI to predict the performance of GSI under various conditions, and we’ve basically asked, ‘Is this GSI functioning well under various conditions, or does it need maintenance soon?’” Dr. Jiao says.

To date the team, with support from a pair of National Science Foundation grants, has collected weather data and geographic features from Philadelphia, Washington, D.C., and Austin, Texas. Dr. Smith is using the data to help answer a foundational question: If a storm hit tomorrow, would a system fail?

“As cities face this increasing burden of how to manage their stormwater effectively and efficiently, we try to take a comprehensive view of the challenge,” Dr. Smith says.

Dr. Jiao’s models, still in their early stages, could be transformative for urban planning because they consider risks alongside income levels on a block-by-block basis. Impoverished communities, for example, could lack plants that slow the movement of floodwaters. Precise data could put planners in a better place to prevent damages with pinpoint interventions.

As with many AI projects, solutions are the result of diverse expert perspectives. “I’m an AI guy, meaning I know the math and I know the algorithms,” Dr. Jiao notes. “But I don’t have insights into socioeconomic factors, and without that we don’t have data to start with.”

That’s where other teammates come in. Besides Dr. Jiao and Dr. Smith, the GSI project’s co-principal investigators include Villanova colleagues Bridget Wadzuk, PhD, of Civil and Environmental Engineering and Peleg Kremer, PhD, of Geography and the Environment in the College of Liberal Arts and
Sciences. The quartet meets weekly or biweekly with sociologists, computer scientists and graduate students to develop insights and solutions into GSI system management.

Even before starting the project, Dr. Smith notes, the team interviewed municipal leaders, community members and engineering firms about stormwater interventions to better inform AI algorithms and create plans for communities.

“We want to create actionable information that the common person can understand,” she says.