HARRISBURG, PA — In July, three graduate students from Harrisburg University of Science & Technology presented their collaborative research at the IEEE 11th International Conference on Big Data Computing Service and Machine Learning Applications (BigDataService 2025), a global event bringing together researchers, professionals, and students in Tucson, Arizona. Their paper, “A Lightweight and Interpretable Ensemble of CNN and Transformer Models for IDC Breast Cancer Classification,” explores how artificial intelligence (AI) can improve the early detection of invasive ductal carcinoma (IDC), the most common and aggressive form of breast cancer.
The research team includes Faith Tobore Edafetanure-Ibeh (Ph.D. Data Sciences, ‘26), Sonachi Bukola Mogbogu (Ph.D. Data Sciences, ‘27), and Ralchukwu N. Mogbogu (MS Analytics, ‘26). Together, they developed a hybrid AI model that combines convolutional neural networks (CNNs) and lightweight transformer models to classify breast tissue images with exceptional accuracy and efficiency.
The group’s ensemble approach integrated EfficientNet-B3, ConvNeXt-Tiny, and Swin-Tiny Transformer models, addressing key challenges in biomedical imaging such as class imbalance and lack of interpretability. To enhance fairness and generalization, they applied focal loss during training, test-time augmentation (TTA) during testing, and Grad-CAM to visualize how the model made its predictions, offering valuable transparency for clinical applications.
Their resulting model achieved 92.94% validation accuracy and a 93% F1-score, outperforming single-model baselines and ranking among the top-performing methods on the dataset. The research contributes to the AI and biomedical communities by providing a lightweight, interpretable, and effective approach for IDC detection that can be applied in real-world clinical settings.
Presenting on an International Stage
For all three students, presenting at IEEE BigDataService 2025 was both validating and inspiring. “It was an exciting experience to share our work on an international stage,” Faith said. “We received valuable feedback, made new connections, and gained confidence in the impact of our research.”
Sonachi described the environment as “a very welcoming space,” where participants freely exchanged ideas and learned from one another’s innovations. “One takeaway was seeing how the field of data science is evolving with innovative ideas. It’s a work of art to see the field continue to grow as more technologies evolve,” she said.
Ralchukwu added that presenting was “rewarding and refreshing,” allowing the team to step outside of the classroom and share their findings with experts in the field. “The best part was the opportunity of technical discussions and insightful feedback, while having our work appreciated by seasoned researchers,” she said.
Research with Real-World Impact
Beyond the recognition and experience, this project reinforced each student’s long-term goals of advancing AI-driven healthcare solutions. “Our research offers a practical and scalable model that could support pathologists in early breast cancer detection,” Faith explained. “By making diagnosis faster, more reliable, and explainable, it has the potential to improve care, especially in resource-limited settings.”
For Sonachi, the experience marked a pivotal moment in her academic journey. “This experience has welcomed me to the world of research and development,” she said. “I’m eager to write more, publish and contribute more, and expand my knowledge to broaden my horizons.”
Ralchukwu echoed that sentiment, noting how the conference helped clarify her career path. “This experience has reinforced my commitment to interdisciplinary research, especially integrating healthcare with technology,” she said. “I’m now more motivated than ever to pursue a career in data science.”
Each student credited Harrisburg University’s emphasis on applied learning, innovation, and mentorship for making this achievement possible. “HU continues to give me the platform to grow,” said Sonachi. “With such a diverse class, I’ve been able to learn from others and stay current in the field.”
Faith expressed deep gratitude for the faculty and mentors at HU who supported their research. “I’m thankful for their guidance and encouragement, which made this work possible,” she said.
Through their work, the three students reflect Harrisburg University’s commitment to hands-on learning and research that makes a meaningful difference in people’s lives.
Want to learn more about Harrisburg University’s Master of Science in Analytics or Ph.D. in Data Science? Visit HarrisburgU.edu or start a conversation with the Admissions Team today.
ABOUT HARRISBURG UNIVERSITY
Harrisburg University was recently recognized by US News & World Report as one of the nation’s most innovative colleges. HU is accredited by the Middle States Commission on Higher Education and is a private, nonprofit university offering bachelor’s and graduate degree programs in science, technology, engineering, mathematics, nursing, and other health science fields. For additional information about HU’s affordable, demand-driven undergraduate and graduate programs, please call 717.901.5146 or email Connect@HarrisburgU.edu. Stay updated by following Harrisburg University on LinkedIn, Instagram, and Facebook.
# # #
