By Eliza Dysart
From veterinary pathology to cancer genomics to surgery, researchers across disciplines gathered Nov. 5 at the UF Health Cancer Institute’s annual Cancer AI Symposium. The event, organized by the Cancer AI Working Group, explored how artificial intelligence is transforming the way researchers and clinicians study, diagnose and treat cancer.
“This event is really about bringing people together,” said Kiley Graim, Ph.D., an assistant professor in the Department of Computer & Information Science and Engineering and a co-chair of the Cancer AI Working Group. “There are so many engineering and AI projects happening across campus, but often people don’t know about each other. We’re trying to build collaborations and get more communication across disciplines.”
The symposium, now in its fourth year, featured three keynote talks, presentations from the Cancer Institute’s AI pilot grant awardees and a student-led hackathon focused on human-centered data science. A poster session allowed researchers from medicine, veterinary science, pathology and physics to present studies that use AI to analyze complex data, improve diagnostic accuracy and personalize treatment.
The complexity of cancer pushes the boundaries of innate human reasoning, said keynote speaker Ali Zarrinpar, M.D., Ph.D., a professor in the Department of Surgery.
“AI does have this promise to help us make our clinical decisions and help us decipher the available data,” he said. “I like to talk about AI as augmenting our intelligence. I don’t want to remove the remove the human aspect, especially when it comes to clinical care.”
Zarrinpar, also a co-chair of the Cancer AI Working Group, highlighted current and potential clinical and operational AI applications, such as Medical Review Board optimization, scheduling efficiency, real-time prediction and decision support, and enhanced diagnostic accuracy and financial efficiency.
In addition to Zarrinpar’s keynote, Graim gave a keynote on disentangling population and disease signals with AI. Zhoumeng Lin, Ph.D., presented on integrating machine learning with nano-tumor database to support cancer nanomedicine development.
Pilot grant awardees Jason Arnold, Ed.D., and Jay Rosen shared their work on an AI-enabled standardized patient avatar for scalable clinician communication training. Rui Yin, Ph.D., another pilot awardee, presented his work on AI-driven identification of m7G RNA modifications in cancer.
For some presenters, AI serves as both a research tool and a personal mission. David Gorlin, a PGY-1 internal medicine resident, developed an AI model to detect blood cancers using flow cytometry, which is a key diagnostic method in blood disorders.
“My mom was diagnosed with T-cell lymphoma before I started medical school, and she had a delayed diagnosis,” Gorlin said. “That inspired me to think, ‘how can I prevent this from happening to other families?’”
Working with 400 cases and 16 biomarkers of chronic lymphocytic leukemia, his model achieved 83% accuracy, mainly built on his own MacBook.
“It started as a garage experiment,” he said. “Now, we think we could develop a real tool clinicians can use.”
Jaclyn Hall, Ph.D., a member of the UF Health Cancer Institute and executive director of the Florida Federal Statistical Data Center, described the vast research possibilities available through the UF-based data center, which allows scientists to link population-level health data from more than 20 federal agencies.
“There’s no other place in Florida where this kind of research can be done,” Hall said. “You can connect vital statistics, mortality data, national surveys, even IRS earnings data. It’s incredibly powerful.”
In the Department of Pathology, Immunology and Laboratory Medicine, Kristianna Fredenburg, M.D., Ph.D., is using AI to make cancer diagnoses more objective. Her research applies a free software tool called QuPath to measure protein expression in tumors.
“As pathologists, we classify tumors by staining them for specific proteins,” Fredenburg explained. “But that process can vary between observers. Teaching AI to recognize staining patterns helps reduce variability and lets us objectively compare results across patients.”
At UF’s College of Veterinary Medicine, graduate student Benjamin Carrier is exploring how AI can assist in identifying biomarkers in canine oral melanoma, a cancer with a poor prognosis for dogs.
“We’re using Halo software to analyze how these biomarkers correlate with survival,” Carrier said. “The goal is to improve how we assess disease and predict outcomes.”
The symposium also featured work that looks beyond human models. Bria Smith, a graduate student in Graim’s lab, presented a data-harmonization pipeline called Paipu that enables large-scale comparative cancer genomics across species.
“Everyone uses mice for research, but mice don’t get spontaneous cancers; they’re induced,” Smith said. “By studying naturally occurring cancers in animals like dogs, we can find better models that may translate more effectively to human disease.”
Meanwhile, Daniela Valdés, a postdoctoral physicist, presented her work on magnetic nanoparticle imaging, using AI-based tools to track how particles move through the body and accumulate in tumors.
“We want to know how long particles stay in the tumor and how they’re distributed in organs,” she said. “AI segmentation helps us analyze that far more efficiently.”
Graim said that variety is what makes the symposium special.
“There are so many cool AI initiatives happening at UF,” she said. “It’s exciting to see them all come together under one roof and imagine how these tools might change cancer care in the years ahead.”
Poster winners:
Zhuochao Huang
“Performance Comparison of Time-Varying Cox, Random Survival Forests, and a Missingness-Aware Neural Network”
Daniela P. Valdés, Ph.D.
“AI-Assisted Quantification of Nanoparticle Biodistribution Using Multi-Modal In Vivo Imaging”
Matthew Chang
“An approach to H-scoring tumors with morphologic heterogeneity using digital pathology and machine learning”
Hackathon winners:
Cancer Patient Diet Cookbook
Team members: Aashish Dhawan, Bushi Xiao, Detravious Brinkley, Chibuzor Okocha, Thien Nguyen, Leena Alabaddane
Github Repository
Artificial Intelligence-derived Small Blood Metabolite Panel Reveals Phenylalanine–Tyrosine Axis Dysregulation Enabling Early Diagnosis of Lung Adenocarcinoma
Team members: Nathan Gilman, Forcha Peter Oben Akem, Qingchen (Jeremy) Yuan, Taylor Tillander
Github repository
