|
Notes:
Notes:
Add code of Single-index varying coefficient model for
functional responses
Regression using imaging responses and some clinical vector
covariate is an important issue in the brain imaging research, and
has drawn a lot of attention in biostatistics and neuroscience
nowadays. One great challenge usually comes from
the smoothness of the function structure, or that for
those voxels close to each other, they have strong
correlations. Motivated by the analysis of a real-diffusion
weighted imaging analysis from the Alzheimer's Disease Neuroimaging
Initiative (ADNI) study, this paper here propose a
nonparametric Single-Index Varying Coefficient model and here
establish this Matlab GUI software. The aim of this software is to
implement a functional analysis
pipeline, for the joint analysis of functional data and clinical
data, for example age, gender and disease status. The model
consists of a nonparametric structure called functional single
index to characterize the association of functional response
as well as a complex spatial-temporal correlation structure
to characterize the proximity and spatially smooth varying
coefficient functions. The software provides
estimation of regression coefficients, nonparametric structure as
well as the spatial-temporal correlation structure.
Furthermore, it can output a simultaneous confidence band for
estimations, and make predictions on any given test set of
design covariate matrix.
Reference
1. Xinchao Luo, Lixing Zhu, Hongtu Zhu. "Single-index varying
coefficient model for functional responses", Annals of Applied
Statistics, Biometrics (2016). Online ISSN: 1541-0420.
Changes:
|
