|Title:||Use of depth measure for multivariate functional data in disease prediction: an application to electrocardiographic signals|
|Date:||Wednesday 13th November 2013|
|Author(s) :||Biasi, R.; Ieva, F.; Paganoni, A.m.; Tarabelloni, N.|
|Abstract:|| In this paper we develop statistical methods to compare two independent samples of multivariate functional data that differ in terms of covariance operators. In particular we generalize the concept of depth measure to this kind of data, exploiting the role of the covariance operators in weighting the components that define the depth.
Two simulation studies are carried out to validate the robustness of the proposed methods.
We present an application to Electrocardiographic (ECG) signals aimed at comparing physiological subjects and patients affected by Left Bundle Branch Block. The proposed depth measures computed on data are then used to perform a nonparametric comparison test among these two populations. They are also introduced into a generalized regression model aimed at classifying the ECG signals.
This report, or a modified version of it, has been also submitted to, or published on
Tarabelloni, N. and Ieva, F. and Biasi, R. and Paganoni, A.M., "Use of Depth Measure for Multivariate Functional Data in Disease Prediction: An Application to Electrocardiograph Signals", The International Journal of Biostatistics, 2015