The University of Texas MD Anderson Cancer Center UTHealth Graduate School of Biomedical Sciences Dissertations and Theses (Open Access)
Author ORCID Identifier
Date of Graduation
Biostatistics, Bioinformatics and Systems Biology
Doctor of Philosophy (PhD)
FURTHER ADVANCES FOR THE SEQUENTIAL MULTIPLE ASSIGNMENT RANDOMIZED TRIAL (SMART)
Tianjiao Dai, M.S.
Advisory Professor: Sanjay Shete, Ph.D.
Sequential multiple assignment randomized trial (SMART) designs have been developed these years for studying adaptive interventions. In my Ph.D. study, I mainly investigate how to further improve SMART designs and optimize the interventions for each individual in the trial. My dissertation has focused on two topics of SMART designs.
1) Developing a novel SMART design that can reduce the cost and side effects associated with the interventions and proposing the corresponding analytic methods. I have developed a time-varying SMART design in which the time of the intervention varies among participants and contain part of the information regarding the intervention effect. We proposed two analytic approaches for analyzing the data from this type of SMART design. Based on simulations, we suggest using joint modeling as a data analysis method since it can well utilize the information of the intervention effect contained in the treatment time and estimate the model parameters better than the single mixed effect model. We also showed that the proposed time-varying SMART design is more efficient than the existing standard SMARTs with respect to the cost and side effects associated with the interventions, while maintaining the same power as the standard SMART design when selecting the optimal embedded adaptive intervention.
2) Developing a new allocation strategy for SMART designs using a response-adaptive, covariate-balanced and optimal-decision-consistent randomization probability under the Bayesian framework. This method applied the existing randomization strategies in clinical trials to SMART designs by accounting for its special framework. In addition, it takes into account the optimization of the individual’s intervention using a Q-learning approach in addition to being response-adaptive and balancing covariates between competing interventions at each SMART stage. This approach also takes advantage of the Bayesian framework. Using simulation studies, we compared the proposed allocation strategy to other possible and existing allocation strategies in clinical trials.
The research on SMART designs I conducted in my Ph.D. study will benefit the community of researchers in the areas of clinical trial design and social behavioral research. The novel design and analysis I proposed will increase the efficiency of SMARTs in terms of the time and cost and reduce the side effects associated with the interventions while promoting a better understanding of the optimal individualized intervention strategy. The new randomization strategy I developed for SMART designs increases the consistency of the optimal intervention strategy for each individual in the trial, which suggests an advantage over other existing randomization methods in clinical trials that can be applied to SMART designs.
SMART, time-varying SMART, joint model, intervention, adaptive, randomization, optimal, Q-learning
Applied Statistics Commons, Biostatistics Commons, Clinical Trials Commons, Design of Experiments and Sample Surveys Commons, Medicine and Health Sciences Commons, Statistical Methodology Commons, Statistical Models Commons