Session SA1.4

Smoothing and Discriminating MAP Data

K Jin*, NL Stockbridge

FDA/CDER
White Oak, MD, USA

The biological data collected from intensive care units contains signal and noise. To extract information that will be useful for predicting, or discriminating the cases likely to develop an acute hypertensive episode (AHE), we begin by applying a spline based smoothing method to the observed MAP curves and related physiological and pathological variables that might contribute to AHE. An empirical smoothing parameter selection will be employed to obtain optimal estimation of signal structure by balancing signal and noise. A cross-validation type method will be applied to training and test datasets to help determine an appropriate model and select relevant variables. From the fitted model, a parameter, or a group of parameters, that best discriminates the H and C cases in the training set will be derived for the prediction of AHE.

(Abstract Control Number: 281)