摘要：Materko W. Stratification Fitness Aerobic Based on Heart RateVariability during Rest by Principal Component Analysis and KmeansClustering. JEPonline 2017;21(1):91-101. The purpose ofthis study was to stratify the degree of aerobic fitness from the heartrate variability in subjects with similar physical and anthropometriccharacteristics by Principal Component Analysis (PCA) and Kmeansclustering. The PCA was used for dimensionality reductionand the initial centroid was computed. Then, it was applied to the Kmeansclustering algorithm for unsupervised learning tasks. Seventytreadmill runners were subjects in this study. After recording theresting tachogram with a cardio frequency meter for 5 min, amaximal cardiopulmonary incremental test was performed tomeasure the maximum oxygen uptake (VO2 max). The powerspectral density (PSD) function of the resting tachograms wasestimated by Welch periodrograms after cubic spline interpolationand resampling at 4 Hz. The PCA was applied to the PSD functionfollowed by the clustering method K-means to obtain two groups with34 (Group 1) and 36 (Group 2) subjects. According to the Student ttest, the cluster of Group 1 presented VO2 max values significantlyhigher than Group 2 (47.1 ± 5.7 vs. 39.3 ± 7.2 mL·kg-1·min-1,respectively; P=0.01). The findings indicate that the proposedmethod appears to be capable of stratifying the degree of aerobicfitness in resting healthy volunteers.
