学术报告（周天剑 2026.6.12）

摘要：In this talk, I will present some methodological contributions to the design and analysis of early-phase oncology clinical trials. I will begin with phase I dose-finding designs aimed at identifying the maximum tolerated dose. In trials with staggered enrollment or late-onset toxicities, I introduce adaptive methods that enable real-time dose-assignment decisions while appropriately accounting for pending toxicity outcomes. I will review statistical frameworks that leverage patients' time-to-toxicity information to guide these decisions, as well as practical considerations for implementation, including the role of suspension rules. This portion of the talk draws from my review article: https://doi.org/10.1214/23-STS895. In the second part of the talk, I will discuss statistical methodology for phase II basket trials, which evaluate a single treatment across multiple cancer types. I will introduce approaches for making efficient go/no-go decisions across cancer types through judicious borrowing of information, thereby improving power while maintaining appropriate control of type I error. This work is based on https://doi.org/10.1093/biomtc/ujag093.