Decision Analysis and Classification Methods for SUDEP Risk Factor Identification

The combination of numerous factors with interactions and small numbers of observations creates two significant challenges for SUDEP research. First, small numbers of observations (n) in the face of relatively large numbers of parameters (p) or risk factors might exceed the practical limits of parametric modeling techniques (National Research Council 2001). Second, the absence of an incontrovertible “gold standard” diagnosis based on attributes of the risk factors adds to uncertainty in the performance of predictive and prognostic models.