Related Experiment Video
Updated: Aug 8, 2026
08:02
Combination Radiotherapy in an Orthotopic Mouse Brain Tumor Model
Published on: March 6, 2012
A multivariate random-effects model with restricted parameters: application to assessing radiation therapy for brain
Hong-Bin Fang1, Guo-Liang Tian, Xiaoping Xiong
1Division of Biostatistics, University of Maryland Greenebaum Cancer Center, 22 South Greene Street, Baltimore, MD 21201, USA. hfang@umm.edu
Statistics in Medicine
|October 13, 2005
Summary
This study introduces restricted multivariate random coefficient models for analyzing longitudinal clinical data with parameter constraints. The method enhances statistical analysis in medical research, particularly for pediatric brain tumor studies.
Area of Science:
- Biostatistics
- Medical Statistics
- Longitudinal Data Analysis
Background:
- Multivariate random-effects models are used for longitudinal clinical data.
- Parameter constraints arise in medical research, such as in pediatric brain tumor studies.
- Existing models may not adequately handle these parameter restrictions.
Purpose of the Study:
- To propose multivariate random coefficient models with restricted parameters.
- To derive maximum likelihood estimates (MLE) for these restricted models.
- To apply the proposed method to analyze pediatric brain tumor radiation therapy data.
Main Methods:
- Development of a class of multivariate random coefficient models with parameter constraints.
- Derivation of maximum likelihood estimates (MLE).
- Application of a modified Expectation-Maximization (EM) algorithm for constrained quadratic optimization.
Main Results:
- The proposed models effectively incorporate parameter constraints in longitudinal data analysis.
- Maximum likelihood estimates (MLE) were successfully derived using the modified EM algorithm.
- The method demonstrated utility in analyzing a pediatric brain tumor study, accounting for tissue relaxation time ordering.
Conclusions:
- Restricted multivariate random coefficient models provide a robust framework for analyzing longitudinal data with inherent parameter constraints.
- The modified EM algorithm is effective for estimating parameters under linear inequality constraints.
- This approach offers improved statistical power and accuracy in specific medical research applications, like pediatric oncology.
