Activity Number:
|
36
- Methods for Cancer Epidemiology
|
Type:
|
Contributed
|
Date/Time:
|
Sunday, July 29, 2018 : 2:00 PM to 3:50 PM
|
Sponsor:
|
Section on Statistics in Epidemiology
|
Abstract #329576
|
|
Title:
|
Evaluating Discriminatory Accuracy of Models Using Partial Risk Scores in Two-Phase Studies
|
Author(s):
|
Parichoy Pal Choudhury* and Anil Chaturvedi and Nilanjan Chatterjee
|
Companies:
|
Johns Hopkins Bloomberg School of Public Health and National Cancer Institute and Johns Hopkins University
|
Keywords:
|
Area under the curve;
Discriminatory accuracy;
Nested case-control study;
Risk-score;
Two-phase studies
|
Abstract:
|
Prior to clinical applications, risk prediction models need to be evaluated in independent studies (e.g., prospective cohort studies) that did not contribute to model development. Often prospective cohort studies ascertain information on some expensive biomarkers in a nested sub-study of the original cohort, typically selected based on case-control status and additional covariates. We propose an efficient approach for evaluating Area Under the Curve (AUC) using data from all individuals irrespective of whether they were sampled in the sub-study. The approach involves estimating probabilities of risk-scores for cases being larger than those in controls conditional on partial risk-scores as opposed to multivariate risk factor profiles. This allows estimation of the underlying conditional probabilities using subjects with complete covariate information in a non-parametric fashion even when numerous covariates are involved. We evaluate finite sample performance of the proposed method and compare it to an inverse probability weighted (IPW) estimator through extensive simulation studies. We apply the method to evaluate a lung cancer risk prediction model using data from the PLCO trial.
|
Authors who are presenting talks have a * after their name.