SUMMARY
Nansu Zong, Ph.D., leads an artificial intelligence and real-world data research program that focuses on developing generalizable computational methods for learning from longitudinal electronic health records (EHRs) and biomedical knowledge sources. His work spans modern natural language processing, biomedical knowledge graphs — including those used in pharmacogenomics — graph-based representation learning and the standardization of clinical data to support scalable evidence generation. These efforts advance clinical research and precision medicine across cardiovascular disease, Alzheimer's disease and related dementias, and cancer.
Focus areas
- Natural language processing and large-scale language model systems for clinical text and protocols. Dr. Zong develops modern natural language processing pipelines that combine large language models with retrieval-augmented generation and established natural language processing techniques to transform unstructured clinical text — such as raw pathology reports — and trial protocols into computable representations. These representations support reliable information extraction and automated reasoning about clinical trial eligibility.
- Biomedical knowledge graphs and pharmacogenomics-informed modeling. Dr. Zong builds and leverages heterogeneous biomedical knowledge graphs that integrate pharmacogenomics, drug-target-disease relationships and multisource biomedical evidence to support mechanistic inference, candidate therapy prioritization and evidence-grounded discovery.
- Graph networks for longitudinal EHR computation. Dr. Zong designs graph-based learning frameworks — such as graph neural networks and temporal or heterogeneous graph structures — to represent patient trajectories, treatment histories and multimodal clinical events. These models advance prediction, patient stratification and the study of treatment effects across complex clinical populations.
- Clinical data standardization and computable phenotyping. Dr. Zong develops workflows that use widely adopted interoperability standards to harmonize EHR data, strengthen the reliability of phenotype definitions and enable portable, reusable analytics.
Significance to patient care
Dr. Zong's research helps patients get the right treatments more quickly and safely. Using advanced computational methods, his team learns from real-world health information — such as electronic health records — to understand how different treatments work for different people. Dr. Zong and his colleagues create tools that can predict who may be at higher risk of certain health conditions, which treatments are most effective for specific patient groups and which patients may benefit from closer monitoring. Their systems also connect patients with clinical trials that fit their medical needs, helping people find new treatments and ensuring that trials enroll the right participants.
Dr. Zong's group uses real-world health data to preview clinical trial plans before they begin. This allows clinical teams to design studies that are realistic, enrollable and more likely to answer important medical questions. They also build tools that give healthcare professionals clearer views of patient histories and treatment options, supporting care plans that reflect each person's unique health journey. Together, this work supports a healthcare system that learns from every patient and delivers more-personalized care
Professional highlights
- Chair, Knowledge Representation Working Group, American Medical Informatics Association, 2025-present.
- Adjunct assistant professor, University of Minnesota, 2023-present.