This article presents a phase I/II trial design for targeted therapies and immunotherapies, with the objective of identifying the optimal dose (OD). We employ a joint modeling technique for discrete time‐to‐event toxicity data and repeated and continuous biomarker measurements. For the biomarker measurements, we implement a change point linear mixed effects skeleton model. This model can accommodate both plateauing and nonplateauing dose‐activity relationships. For each new cohort of patients, we estimate the maximum tolerated dose (MTD) taking toxicity as a cumulative endpoint, over six treatment cycles. Then, we select the OD using two different criteria. The OD is a dose that is equally active to the MTD or a dose located on the beginning of the plateau of the dose‐activity relationship. Joint modeling allows us to take into account informative censoring due to toxicities or lack of activity and we also consider consent withdrawal and intermittent missing responses. Model estimation relies on likelihood inference.

中文翻译：

---

在早期试验中用于识别活动平台的变化点联合模型

本文介绍了针对靶向疗法和免疫疗法的I / II期试验设计，目的是确定最佳剂量（OD）。我们采用联合建模技术来处理离散的事件毒性数据以及重复和连续的生物标志物测量。对于生物标志物的测量，我们实现了一个变化点线性混合效应骨架模型。该模型可以适应平稳和非平稳剂量-活性关系。对于每个新的患者群体，我们估计在六个治疗周期中以毒性作为累积终点的最大耐受剂量（MTD）。然后，我们使用两个不同的标准选择OD。OD是对MTD具有同等活性的剂量，或者是位于剂量-活性关系的平稳期开始时的剂量。联合建模使我们可以考虑由于毒性或缺乏活动而提供的信息审查，并且我们还考虑撤回同意和间歇性缺失响应。模型估计依赖于似然推断。