Nansu Zong, Ph.D., is a renowned expert in the field of artificial intelligence (AI). Dr. Zong's lab, within Mayo Clinic's Department of Artificial Intelligence and Informatics, leverages AI and biomedical knowledge to create advanced AI-based predictive models. His research team is at the forefront of developing cutting-edge AI techniques for computational drug repurposing and predicting treatment outcomes using electronic health records (EHRs).
The ultimate goal for Dr. Zong's team is to create AI-empowered drug development tools and clinical decision support systems for precision medicine in conditions such as cancers, Alzheimer's disease and cardiovascular diseases. To achieve this goal, his team is using advanced AI techniques, such as graph neural networks, embedding, generated models and reinforcement learning, to process biomedical knowledge graphs and EHR data. Specifically, the team is interested in leveraging AI techniques to develop precision medicine approaches to improve patient outcomes and reduce health care costs.
Under Dr. Zong's leadership, his lab is advancing the field of AI and biomedical research by developing new AI-based tools that can assist in drug development and precision medicine. With their innovative and cutting-edge research, Dr. Zong and his team are helping to transform health care by creating new ways to predict disease outcomes and develop personalized treatment plans for patients.
- Computational drug repurposing. Dr. Zong's team is pioneering the development of novel AI methods for computational drug repurposing. This involves identifying new uses for existing drugs by leveraging AI techniques to analyze large volumes of biomedical data, including:
These techniques are employed to process biomedical knowledge graphs that identify patterns and relationships that can be used to predict drug efficacy and side effects. This innovative research is helping to accelerate drug development, identify new treatments for diseases with limited treatment options and get treatments options to patients more quickly.
- Graph neural networks, which can include variants such as graph convolutional networks.
- Embedding, such as graph embedding.
- Large language models, using transformers for example.
- Clinical decision support. This includes prediction of disease and treatment outcomes, as well as optimal regimen learning. Dr. Zong's team is developing clinical decision support systems based on EHRs. This involves using AI techniques, such as reinforcement learning and large language models, to analyze large volumes of patient data to predict treatment outcomes and identify potential risks and complications. His team also is developing tools to support clinical decision-making by providing real-time alerts to health care teams and recommendations for optimal treatment plans. By leveraging AI techniques, he and his colleagues aim to develop personalized treatment plans that consider a patient's unique medical history and genetic makeup.
- Deep phenotyping based on EHRs. Deep phenotyping is the process of extracting detailed and comprehensive clinical data from EHRs to develop a more complete understanding of a patient's medical history and condition. Dr. Zong's team is leveraging AI techniques for deep phenotyping based on EHR data. The team's approach involves modeling large volumes of this data, including structured medication and lab test data and unstructured patient clinical notes, to feed advanced deep learning models. Their aim is to develop a more complete understanding of a patient's medical history and condition. Then provide a feasible data-driven tool that can automatically learn the subcohort of the patients to identify subtypes. Deep phenotyping can aid informed treatment decisions and personalized care. For example, this study focused on cancers, Alzheimer's disease and cardiovascular diseases.
Significance to patient care
Dr. Zong's work applying AI techniques to health care has significant implications for patient care. By developing AI-driven predictive models using health care data and biomedical knowledge, his team is pioneering the development of novel methods for computational drug repurposing, clinical decision support and deep phenotyping using EHRs.
Dr. Zong's team also is developing computational drug repurposing methods by analyzing data from biomedical knowledge graphs. By identifying existing drugs that may be effective for conditions with limited treatment options, this approach to drug development can provide new treatments, reduce health care costs and improve overall patient outcomes.
Additionally, the development of AI-driven predictive models has the potential to identify the most effective treatments for a given condition, leading to more personalized and effective care and improved outcomes. By leveraging vast amounts of EHR data, Dr. Zong's team can develop highly accurate and effective predictive models to help health care teams make informed decisions about the best treatments for their patients. These models can help lower health care costs by identifying the most efficient and cost-effective treatments.
Overall, Dr. Zong's team is working to identify the most effective treatments and minimize potential risks and complications. Their innovative research based on cutting-edge AI technologies is helping to advance the field of drug development and precision medicine. This research can further transform health care by presenting health care workers with the tools they need to make informed decisions and provide the best possible care.
- Vice chair, Biomedical Knowledge Representation and Semantics Working Group, American Medical Informatics Association, 2023.