The research of Dennis H. Murphree Jr., Ph.D., centers on the intersection of artificial intelligence (AI) and medicine, where he focuses on applications of machine learning and computer vision to clinical questions rather than those of basic science. Dr. Murphree utilizes a broad portfolio of approaches, including classical linear and nonlinear predictive modeling, as well as deep learning-driven image analysis to better understand both the current and future state of the patient.
Dr. Murphree aims to discover new diagnostic and risk prediction methods and translate them to clinical practice. He is particularly interested in applications in cardiovascular medicine and in dermatology. He is also interested in integrative approaches that fuse data from multiple domains, for example imaging and genomic data.
Dr. Murphree currently serves as co-director of the Office of Artificial Intelligence in Dermatology at Mayo Clinic.
- Multimodal artificial intelligence for dermatologic cancer. Dr. Murphree utilizes cutting-edge methods from machine learning and computer vision to address a variety of skin cancers, particularly melanoma and cutaneous squamous cell carcinoma. These studies integrate image data, gene expression and genomic data, as well as traditional electronic health record data to diagnose disease, stratify risk and identify candidates for adjuvant therapy.
- Machine learning for improved medical risk prediction. Dr. Murphree and his colleagues, notably Gyorgy Simon, Ph.D., at the University of Minnesota, have a comprehensive research program focused on predicting a variety of postoperative complications. One novel aspect of this research is the direct incorporation of multi-institutional data in a variety of federated learning approaches, allowing for improved performance and generalizability while maintaining data privacy.
- Automated analysis of electrocardiogram (ECG) via deep learning. Along with other members of the Office of Artificial Intelligence in Cardiovascular Medicine, Dr. Murphree studies the use of deep learning methods to learn and predict a variety of outcomes from ECG data, including screening for cardiac amyloidosis. He is also involved in some early experimental studies applying AI methods to positron emission tomography.
- Practical model translation. Dr. Murphree is keenly interested in translating models from the research arena to practical clinical use. He has been heavily involved in the deployment of several machine learning models to active clinical practice.
Significance to patient care
Dr. Murphree's work directly improves patient care by incorporating methods from AI and data science into clinical research and practice. For example, more accurately predicting the likelihood of melanoma metastasis might help patients avoid unnecessary surgery, or else begin appropriate treatment regimens sooner. AI-enhanced ECG screening for cardiac amyloidosis might promote early diagnosis, potentially leading to better outcomes. More generally, Dr. Murphree hopes that by developing AI tools that augment the clinical team's human intelligence his work will improve care delivery for all.
- Early Career Reviewer, National Institute of Health, 2019
- J.W. Gibbs Fellow, Yale University Department of Physics, 2002
- Fulbright Fellow, U.S. Department of State, 2001