American Statistical Association New York City Metropolitan Area Chapter Memorial Sloan-Kettering Cancer Center Biostatistics Seminar

This talk introduces a class of point impact regression models in which the trajectory of a continuous stochastic process, when evaluated at one or more locations, is associated with a scalar response. A motivation for developing this type of model arises from genome-wide expression studies in which it is of interest to locate genes associated with clinical outcomes. The observed trajectories are assumed to have fractal properties (fractional Brownian motion) in the neighborhood of the targeted location. Bootstrap confidence intervals for these points are developed. Non-Gaussian limit distributions and faster-than-root-n rates of convergence that depend on the Hurst exponent of the fractional Brownian motion play a central role.