Kean University Researchers Discover Plant Fat and Animal Protein May Help Prevent Type 2 Diabetes

For years, researchers have explored a range of possible causes for Type 2 diabetes, looking at everything from lifestyle to diet and beyond. The key challenge has been disentangling these overlapping factors to determine which ones actually contribute to the disease.  

Now, new machine learning techniques have helped two Kean University researchers cut through the noise and pinpoint the most influential factors. Their findings, published in the Journal of Clinical Medicine, revealed that a combination of age, body mass index, fatty acid ratios and an unhealthy lifestyle are linked to Type 2 diabetes.  

In their analysis, co-authors Adam Eckart, DHSc, and Pragya Sharma, Ph.D., both assistant professors in Kean’s Department of Health and Human Performance, uncovered that higher protein intake from animal sources and plant fat were associated with a lower risk of Type 2 diabetes.  

“Typically, broad recommendations for individuals with Type 2 diabetes emphasize physical activity. However, we must also consider the types of foods we’re eating,” Sharma said. “Our research highlights the importance of animal protein and plant-based fats, like olive oil and avocados, and aims to refine dietary perspectives and prevention strategies.”  

Eckart and Sharma used a machine-learning technique called XGBoost, a specialized form of predictive modeling that helps analyze complex data and predict outcomes. The co-researchers analyzed data from the National Health and Nutrition Examination Survey and were able to address the issue of confounding variables. For example, people who eat animal fats might also have unhealthy lifestyles, so it’s difficult to determine which factor(s) contribute to Type 2 diabetes.  

“What makes this study unique is our use of a relatively new machine-learning technique that uses decision trees – a series of yes/no binaries that help predict outcomes,” Eckart said. “It helps us to analyze health outcomes with more than one predictor. Tools like XGBoost make it as practical as possible to identify dietary and lifestyle variables and cut through the interdependencies.” 

The innovative approach exemplifies the cutting-edge research happening at Kean, which recently earned a prestigious R2 research designation from the Carnegie Classification of Institutions of Higher Education for its research and doctorate production. 

“This novel research by Drs. Eckart and Sharma is a great example of how Kean scholars are using machine learning in the analysis of large data sets,” said James F. Konopack, Ph.D., dean of Kean’s College of Health Professions and Human Services. “Tools like XGBoost are helping us more completely understand the complex interactions among disease predictors and improve public health.” 

Diabetes, a chronic metabolic disease characterized by elevated blood-sugar levels, affects more than 830 million people worldwide –– a number that has steadily risen, according to the World Health Organization. In the United States, the American Diabetes Association reports, most of the 38 million people with diabetes have Type 2, which most often develops in adults. 

"I am thrilled at our faculty’s recent publication in the Journal of Clinical Medicine,” said Consuelo Bonillas, Ph.D., chair of the Department of Health and Human Performance and professor of public health. “It is a wonderful example of how our research at Kean can help alleviate the burden of chronic disease with practical recommendations that people can use in their everyday lives."