Title: Nonparametric depth based methods for analyzing health data
Authors: Sara Lopez Pintado - Northeastern University (United States) [presenting]
Abstract: Technological development in many emerging research fields has led to the acquisition of large collections of data of extraordinary complexity. In neuroscience for example, brain-imaging technology has provided us with complex collections of signals from individuals in different neurophysiological states in healthy and diseased populations. The development of statistical tools to analyze this type of high-dimensional data sets is very much needed. New robust methodologies for analyzing functional and imaging data based on the concept of depth are presented. Functional depth provides a rigorous way of ranking from center-outward a sample of functions. This ordering allows the definition of robust descriptive statistics such as medians, trimmed means and central regions for functional data. Moreover, data depth is often used as a building block for developing robust statistical methods and outlier-detection techniques. Permutation depth-based tests for comparing the location and dispersion of two groups of functions or images are proposed and calibrated. The performances of these methods are illustrated in simulated and real data sets. In particular, we tested differences between brain images of healthy controls and patients with severe depression. We also used these methods to analyze differences between the growth pattern of normal and obese children.