Focusing on methodological development of machine learning for individualized treatment rules in precision medicine
Precision medicine aims to deliver the right treatment to the right patient at the right time. To achieve this, clinicians and researchers benefit from the use of treatment models developed for the individual. We combine our group’s research and collaborative clinical work to advance the study and application of precision medicine in clinical research. The main objective of our work is to develop individualized treatment rules (ITRs) to help guide patients to the correct treatment that produces the best therapeutic effects.
Our group assists the study and advancement of precision medicine through our methodology development and clinical research support.
Method and ITR development
The major focus of our methodology work is the development of new biostatistical learning algorithms in precision medicine. We work on the methodological development of machine learning for the derivation of individualized treatment rules (ITRs) to tailor treatments for patients with heterogeneous responses to different medications and health care practices. The main goal is to find a treatment rule that will provide patients an individualized treatment that optimizes therapeutic effects.
We are also currently involved in an international clinical trial that compares different treatments for spinal muscular atrophy. A portion of the trial’s data lacks patient-specific information, so we create new methods to incorporate summary statistics from this limited dataset. By developing new methods, we can help derive a better division function to determine an effective precision medicine approach for this cohort.
Beyond these two areas within method development, our group participates in longitudinal and spatio-temporal data analyses.
Clinical research support
Outside our research in method development, there is an interest in clinical projects related to maternal health, environmental health, infectious diseases, and transplantation. We currently have a collaboration with the Department of Bone Marrow Transplantation and Cellular Therapy in which we build predictive machine learning models—using established machine learning algorithms—that predict the overall survival for patients receiving bone marrow transplantation at St. Jude.
We are also involved in an ongoing St. Jude clinical trial focused on lymphoma, in which we monitor the trial design and develop a robust statistical analysis plan.
Through our involvement in both the development of novel machine learning algorithms and the support of clinical research, our contribution to the application of precision medicine helps advance clinical research and care.
Dr. Yiwang Zhou is an Assistant Member of the St. Jude Faculty and received her PhD in Biostatistics from the University of Michigan. As a Faculty member at St. Jude, Dr. Zhou continues her research in the development and improvement of individualized treatment rules (ITRs) for use in personalized medicine. Her proposed statistical learning methods have been previously applied to clinical trials, and her collaborations extend to numerous diverse projects. Dr. Zhou’s training, expertise, leadership, and motivation continually lead to the successful implementation of academic research projects.