ppseq: An R Package for Sequential Predictive Probability Monitoring

Advances in drug discovery have produced numerous biomarker-guided therapeutic strate-gies for treating cancer. Yet the promise of precision medicine comes with the cost of increased complexity. Recent trials of targeted treatments have included expansion cohorts with sample sizes far exceeding those in traditional early phase trials of chemotherapeutic agents. The enlarged sample sizes raise ethical concerns for patients who enroll in clinical trials, and emphasize the need for rigorous statistical designs to ensure that trials can stop early for futility while maintaining traditional control of type I error and power. The R package ppseq provides a framework for designing early phase clinical trials of binary endpoints using sequential futility monitoring based on Bayesian predictive probability. Trial designs can be compared using interactive plots and selected based on measures of efficiency or accuracy.