Analysis and visualization of health data

Abstract

Health data is important and available online. We have studied visualization of two kinds of datasets. First is a survey data from Clinical, Anthropometric & Bio chemical Survey data (CAB) dataset and second is medical records of individuals from PhysioBank. With CAB dataset we do binary classi cation of data. Often one would like to subset the data set according to the value of such a binary variables and compare and contrast the statistical parameters for such subsets. We have written an R code to analyze such datasets and create plots to give comparison of mean of variables for such binary partitions. We show the result of this analysis for the CAB database which has health data from several Indian states. The data contains survey from the year 2014 with total 52 health indicators, covering 8 states. We also show the state-wise means of several partitioned and normalized variables using a single plot. Twobinary variables have been used, a demographic one (rural/urban) and the gender (male/female), to partition and compare the database. One of the major problem is understanding heart issues and their prediction. For the second dataset we analyze heartbeat signals for normal and arrhythmia data of individual records. We have used MIT-BIH Arrhythmia Database (MITDB)and MIT-BIH Normal Sinus Rhythm Database(NSRDB) to understand heart issues. We visualize the heart rhythm data using phase plot and rst return map to see the di erence in normal and arrhythmia data