Biostatistics and Computer-based Analysis of Health Data using R

Christophe Lalanne, Mounir Mesbah

• Publisher: ISTE Press - Elsevier
• Year: 2016
• Print ISBN: 9781785480881
• eBook ISBN: 9780081011751
• Topic: Computer Science
• Subtopic: Business Intelligence
• Index: Computer Science

Biostatistics and Computer-Based Analysis of Health Data Using the R Software addresses the concept that many of the actions performed by statistical software comes back to the handling, manipulation, or even transformation of digital data. It is therefore of primary importance to understand how statistical data is displayed and how it can be exploited by software such as R. In this book, the authors explore basic and variable commands, sample comparisons, analysis of variance, epidemiological studies, and censored data. With proposed applications and examples of commands following each chapter, this book allows readers to apply advanced statistical concepts to their own data and software. - Features useful commands for describing a data table composed made up of quantitative and qualitative variables - Includes measures of association encountered in epidemiological studies, odds ratio, relative risk, and prevalence - Presents an analysis of censored data, the key main tests associated with the construction of a survival curve (log-rank test or Wilcoxon), and the Cox regression model

1

Elements of the Language

Abstract:

R is more than a simple software program for statistics; it is a language for the manipulation of statistical data. This partly explains its difficult non user-friendly approach for users accustomed to drop-down menus such as those offered by SPSS (although SPSS also offers a basic macro language). This chapter allows the reader to discover the elements of the language and to become familiarized with the mechanisms by which to represent statistical data in R. In the illustrations that follow, the R commands are prefixed with the symbol >, which designates the R console prompt. It is, therefore, unnecessary to copy this symbol to test the proposed instructions.

Keywords

Birthwt data; Categorical variables; Data representation; Data table; Multidimensional data management; Numerical variables; Observations; Packages; R scripts; R Studio

R is more than a simple software program for statistics; it is a language for the manipulation of statistical data [IHA 96, VEN 02]. This partly explains its difficult non user-friendly approach for users accustomed to drop-down menus such as those offered by SPSS (although SPSS also offers a basic macro language). This chapter allows the reader to discover the elements of the language and to become familiarized with the mechanisms by which to represent statistical data in R. In the illustrations that follow, the R commands are prefixed with the symbol > , which designates the R console prompt. It is, therefore, unnecessary to copy this symbol to test the proposed instructions.

1.1 Before proceeding

1.1.1 Installing R

The installation of R is relatively easy and instructions can be found at the following website: http://cran.r-project.org. The software program is available for Windows, Linux and Mac. In addition to the R program, the installer provides an R script editor, online help and a set of base packages. The packages include commands specific to a certain area (graphics commands, modeling commands, etc.) and make it possible to enhance the base features of R.

1.1.2 RStudio

Although R is sufficient to start or perform statistical analyses, RStudio (www.rstudio.com) provides a particularly enjoyable working environment for R. It includes a powerful R script editor, a console in which the user can execute the commands (or send those inserted in the script editor), online help, a graphics browser, a data viewer and much more [VER 11]. A brief introduction to the software is provided in the appendices.

1.1.3 List of useful packages

Although the R base packages allow for most statistical analyses addressed in this book, a number of additional packages must be installed to reproduce some of the proposed applications. In addition to the base packages that are used throughout this book (lattice, foreign, survival), etc. the following packages can be installed by means of the install.packages() command, in the format install.packages(“reshape2”): reshape2, gridExtra, vcd (Chapter 3); epiR, car (Chapter 4); ppcor, psych (Chapter 5); epicalc, ROCR (Chapter 6).

These packages can be installed directly from RStudio using the package manager. Before i...

Table of contents

• Cover image
• Title page
• Table of Contents
• Copyright
• Introduction
• 1: Elements of the Language
• 2: Descriptive Statistics and Estimation
• 3: Measures and Tests of Association Between Two Variables
• 4: Analysis of Variance and Experimental Design
• 5: Correlation and Linear Regression
• 6: Measures of Association in Epidemiology and Logistic Regression
• 7: Survival Data Analysis
• Appendix 1: Introduction to RStudio
• Appendix 2: Graphs with the Lattice Package
• Appendix 3: The Hmisc and rms Packages
• Bibliography
• Index