Clustering analysis of vital signs measured during kidney dialysis

Abstract

Analysis of vital data of kidney dialysis patients is presented. The analysis is based on some vital signs of pulse rate (PR), respiration rate (RR) and body movement (BM) which were obtained by a sleep monitoring system. In a series of experiments, eight patients of different genders and ages were involved. For the analysis, a hierarchical clustering method was applied with multi-dimensional dynamic time warping distance to analyze the similarity between the vital signs. The hierarchical clustering uses Ward’s method to calculate the distance between two clusters. The analysis results show that daily vital sign indicates a feature related to one of the clusters and physiological rhythms based on a series of the features vary depending on the season. Based on the hypothesis, some irregular vital signs which deviate from the physiological rhythms can be detected to predict abnormal health condition and discomfort of the patients.