Dr. Casey Cazer, Cornell ’16, ’20, a veterinarian and assistant professor in the Department of Population Medicine and Diagnostic Sciences, has mentored three MPH students in applied research projects that use veterinary clinical data and machine learning to explore antimicrobial resistance (AMR). Using a One Health perspective, they focus on how resistant pathogens cross between human and animal populations, especially pets. She wants her research to help promote more sustainable use of antibiotics in small-animal veterinary medicine.

“When people think about antibiotic use in animals, they often think of livestock,” Cazer says, but while more antibiotics are used in animal agriculture, “we also give them to our pets, and most humans have more contact with pets than with livestock.” The antibiotics used with small animals also tend to be more critical for human medicine. According to MPH student Mu Jin, who created a scoping review for evidence of transmission of bacteria resistant to antimicrobials between humans and pets, “half the U.S. population keeps at least one pet, but there is no regulation of antibiotic usage in pets.”

Most of Cazer’s work is focused on using new computational and surveillance methods. Students in her lab use machine learning and statistical modeling to understand AMR trends, especially in dogs and cats. MPH student Yufan Yang worked with Cazer and MPH associate professor Dr. Kevin Cummings, Cornell ’96, ’10, to analyze resistant infections in cats and correlations between AMR and the introduction of new antimicrobials, based on samples tested at Cornell’s Animal Health Diagnostic Center.

Machine learning is a relatively new method of data analysis, and for MPH student Ning Zhang, whose research uncovers multidrug resistance patterns among dogs, “combining new methods to discover new things about AMR has been really exciting.” However, Cazer cautions, “machine learning is just one approach to process large amounts of data,” and there is research pointing to false conclusions and bias in machine-learning algorithms. She suggests that considering expert opinion in conjunction with these algorithms can help to manage those issues.

Cazer says it has been “eye-opening” to work with MPH faculty and students, especially to see how public health approaches the social-behavioral aspects of health. “I would love to bring more of this perspective into veterinary medicine,” she says, perhaps to understand more about the human-animal bond, and how it might impact health outcomes. When processing large amounts of data using algorithms, it’s important to also leverage expert understanding to interpret results and inform public policy decisions.

Written by Audrey Baker