Study
PLCO
(Learn more about this study)
Project ID
PLCO-948
Initial CDAS Request Approval
Mar 22, 2022
Title
Statistical Inference for Survival Cure Models Using PLCO as Auxiliary Data
Summary
The Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial is a randomized, controlled trial to study impacts of various screening exams on reducing mortality from prostate, lung, colorectal and ovarian cancer. PLCO consisted of about 155,000 participants enrolled between November 1993 and July 2001. Together with its detailed demographic and clinical information, and over 10 years of follow-up, the PLCO cohort can be a unique resource to conduct genetic studies of many cancer-related traits. Our project focuses on the statistical inference and applications of semi-parametric survival models for PLCO survival data using auxiliary information from real world data. We will propose more effective non-parametric and semi-parametric estimation methods by synthesizing external auxiliary information to improve estimation efficiency within small studies of limited sample sizes. With the improvement of medical science, survival data with a cure fraction gradually emerge. We focus on the class of semiparametric cure models whose nonparametric components follow specific shape constraints necessary in applications. We will conduct PLCO data analysis in our project, which contributes to statistical inference and applications for survival data using auxiliary information. We will focus on the following four types of cancers: prostate, colorectal, breast, and lung.
Aims
1. Filling the research gap of studying a small internal data with external data resources.
2. By making full use of rich real world data resources in the big data era, we develop the methods of information synthesis in which external auxiliary information including survival probabilities, parametric estimator values or high-dimensional data is incorporated in small studies.
3. We propose new estimations by the empirical likelihood theory, generalized moment estimation theory and sufficient dimension reduction methods for semiparametric cure models under censored data, and establish the asymptotic properties of semiparametric estimators by the empirical process theory.
4. When the external data and internal data are heterogeneous or the external data is uncertain, the necessary improvement method will be introduced.
Collaborators
Yi Niu, Ph.D., Associate Professor, Dalian University of Technology
Hui Song, Ph.D., Assistant Professor, Dalian University of Technology
