Michigan Biostatistics Presents: Yingqi Zhao, PhD скачать в хорошем качестве

Michigan Biostatistics Presents: Yingqi Zhao, PhD

Presenter: Yingqi Zhao, PhD Professor, Biostatistics Program Fred Hutchinson Cancer Center Presentation Date: November 13, 2025 Title: Addressing Real-world Challenges in Cancer Early Detection and Therapeutic Studies Abstract: This talk highlights two recent developments to address key challenges in cancer early detection and therapeutic studies. First, designing a randomized controlled trial to demonstrate the clinical utility of an early detection biomarker with mortality and related endpoints poses unique challenges. The hurdles stem from the prolonged natural progression of the disease and the lack of information regarding the time-varying screening effect on the targeted asymptomatic population. To address these issues, we offer a new approach to cancer screening trial design, focusing on an alternative endpoint: reducing the incidence of late-stage cancer. This approach could make screening trials more feasible with a shorter follow-up period, and accelerate the translation of findings into public health applications. A five-state disease progression model is leveraged to derive model-based effect sizes to inform the design. We use numerical examples from the National Lung Screening Trial to demonstrate the method. In the second part, we address the challenge of estimating long-term treatment effects by linking clinical trial data with observational follow-up datasets, where linkages are often incomplete. Using the Cox model to define long-term effects, we propose several estimation strategies, including the non-linked-as-censored method, an inverse probability of linkage weighting (IPLW) approach, and an augmented inverse probability of linkage weighting (AIPLW) method. These approaches account for both censoring and incomplete linkage. Simulation results confirm the validity of our method and we apply our methods to the SWOG study. Together, these advances aim to strengthen the design and analysis of cancer studies across early detection and treatment domains.