Pattern Recognition of Cardiovascular and Psychomotor Variability in Response to Pharmacological Agents.

摘要

The goal of this project is the development of pattern recognition and signal processing methods that will provide indices of responsivity to challenge when applied to Army supplied human cardiovascular and psychomotor data. Time series and point process techniques will form the basis of the approach, and the assumptions that underlie the methods will be examined and tested. The relationship of infrequent and brief events, if any, to the indices will be elucidated. This report presents the results of the work over the past year, which has proceeded along three parallel lines: the design, implementation, and testing of data preprocessing steps that restore physiologic integrity to noise corrupted data; the preliminary implementation of evaluation of several clustering and pattern recognition methods; and the selection of a data segmentation algorithm for the partitioning of time series data. The work followed naturally from that of the previous year, in which we reviewed the state of the art of the understanding of the links between the noninvasive measurements described here, and the underlying physiology. Plans are described for the third year of the work, which will combine those separate tasks into a single tool for physiologic state characterization. (jhd)