Compact representation of autonomic stimulation on cardiorespiratory signals by principal component analysis

Multiparametric monitoring of patients allows a better comprehension of their clinical evolution, but yields a large amount of data, difficult to be analysed and compared: this makes desirable a compact data interpretation and representation. The authors describe the application of principal component analysis (PCA), a technique allowing the reduction of the data set dimensionality, to a series of parameters extracted from cardiovascular (ECG, systemic arterial pressure) and respiratory signals. An x-y plot, built up with the first two principal components (PC's), provides a compact representation of the beat-to-beat variation of the signal features as compared with basal conditions, during different autonomic stimulations (passive tilt test Valsalva manoeuvre, handgrip test baroreflex stimulation by phenylephrine administration).<>