Accompanying R code to replicate simulation results for the "Bayesian and Frequentist approaches to sequential monitoring for futility in oncology basket trials: A comparison of Simon’s two-stage design and Bayesian predictive probability monitoring with information sharing across baskets" manuscript to appear in PLOS ONE. Co-authors include myself (Alex Kaizer: alex.kaizer@cuanschutz.edu), Emily Zabor, Lei Nie, and Brian Hobbs.
Briefly, Bayesian predictive probabilities (PP) can be used for sequential monitoring in clinical trials. Specifically, in this manuscript, simulation studies of basket trials with 10 one-arm baskets with binary outcomes are implemented. Comparisons are made between Simon's two-stage minimax design, a Bayesian approach with PP for continual futility monitoring, and a Bayesian approach that further considers information sharing across baskets with multi-source exchangeability models (MEMs).
function_mem_seq_bt.R: a set of functions for implementing the simulation study
probability_threshold_calibration.R: code to calibrate the posterior probability thresholds for the three Bayesian approaches assuming no interim monitoring for the global null scenario
snowfall_sim_memseqbt.R: code to run simulations with parallelization
seqPP_result_code.R: code to take simulation results and summarize with tables and figures
mem_seq_bt_results_part1.txt: text file of simulation results (part 1 of 4, need to merge together with other parts for summarizing results using seqPP_result_code.R)
mem_seq_bt_results_part2.txt: text file of simulation results (part 2 of 4, need to merge together with other parts for summarizing results using seqPP_result_code.R)
mem_seq_bt_results_part3.txt: text file of simulation results (part 3 of 4, need to merge together with other parts for summarizing results using seqPP_result_code.R)
mem_seq_bt_results_part4.txt: text file of simulation results (part 4 of 4, need to merge together with other parts for summarizing results using seqPP_result_code.R)
Seq_PP_ENAR_2022_Presentation.pdf: presentation slides from the ENAR 2022 Meeting relating to the project/paper