|
Activity Number:
|
417
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #309223 |
|
Title:
|
Double Nonparametric Estimation for Relating Dose Distributions to Scalar Outcomes
|
|
Author(s):
|
Matthew Schipper*+ and Jeremy Taylor
|
|
Companies:
|
University of Michigan and University of Michigan
|
|
Address:
|
200 Zina Pitcher Place, Ann Arbor, MI, 48109,
|
|
Keywords:
|
functional data ; monotonicity ; nonparametric regression ; dose effect ; normal tissue complications
|
|
Abstract:
|
Normal tissue complications are a common side effect of radiation therapy. They are the consequence of the non-uniform dose of radiation received by the normal tissue near the tumor. Dose-damage-injury models relate the dose distribution received by the normal tissue to the observed injury/complication in two steps. The first step relates dose to the unobserved damage and the second relates damage to the observed injury. In our model, a summary measure of damage is obtained by integrating a weighting function of dose (W(d)) over the dose density. Similar to a generalized additive model, the linear predictor in our model includes a nonparametric function of damage, H(damage) which relates damage to injury. Both W(.) and H(.) are written as regression splines and estimated nonparametrically and monotonically. We illustrate our method with data from a head and neck cancer study.
|
