Data do not always obey the normality assumption, and outliers can have dramatic impacts on the quality of the least squares methods. We use Huber's loss function in developing robust methods for time-course multivariate responses. We use spline basis expansion of the time-varying regression...