The open-source R package surveillance implements statistical methods for the modeling and monitoring of epidemic phenomena based on (infectious disease) surveillance data. This includes time series of counts, proportions and categorical data as well as spatio-temporal point processes. Potential users are biostatisticians, epidemiologists and others working in, e.g., applied infectious disease epidemiology. However, applications could just as well originate from environmetrics, reliability engineering, econometrics or the social sciences.
Salmon et
al. (2016) provide an overall guide to the monitoring capabilities
of surveillance. The paper is available as vignette("monitoringCounts")
with the package. Further descriptions can be found in a book chapter by
Höhle and Mazick (2010, preprint),
and – slightly outdated – Höhle (2007) or vignette("surveillance").
Aberration detection in count data time series, e.g.,
farringtonFlexible().
Online change-point detection in categorical time
series, e.g., categoricalCUSUM().
A Markov Chain approximation for computing the run-length distribution
of the proposed likelihood ratio CUSUMs is available as function
LRCUSUM.runlength().
See the online reference index for the complete list of algorithms.
Backprojection methods: backprojNP()
Adjusting for occurred-but-not-yet-reported events:
nowcast(), bodaDelay()
Meyer et al. (2017) provide a guide to the spatio-temporal modeling capabilities of surveillance. These so-called endemic-epidemic models have proven useful in a wide range of applications, also beyond epidemiology. A list of corresponding publications is maintained at https://surveillance.R-forge.R-project.org/applications_EE.html.
twinstim()vignette("twinstim")"epidataCS", which holds the
observed events (with covariates) and exogenous covariates on a
space-time grid (for the endemic/background component)epitest() for space-time
interactiontwinSIR()vignette("twinSIR")"epidata"hhh4()vignette("hhh4_spacetime")
for areal time series, and more generally in vignette("hhh4"),
including the univariate case"sts" (see below)"sts"The S4 class "sts" (surveillance time series), created
via sts() or linelist2sts(), represents
(multivariate) time series of counts. For areal time series, the class
can also capture population fractions, a map, and a weight matrix.
For evaluation purposes, the package contains several datasets derived from the SurvStat@RKI database maintained by the Robert Koch Institute in Germany. See the online reference index for the complete list of datasets.
The stable release version of surveillance is hosted on the Comprehensive R Archive Network (CRAN) at https://CRAN.R-project.org/package=surveillance and can be installed via
install.packages("surveillance")The development version of surveillance is hosted on R-Forge at https://R-Forge.R-project.org/projects/surveillance/ in a Subversion (SVN) repository. It can be installed via
install.packages("surveillance", repos = "https://R-Forge.R-project.org")Alternatively, a development build can be installed from the R-universe mirror of R-Forge.
Contributions are welcome! Please report bugs via e-mail to
maintainer("surveillance").
Note that (large) new features are unlikely to be included in surveillance. Some extensions have already been developed in separate packages, for example hhh4contacts, HIDDA.forecasting, hhh4addon, and hhh4ZI.
The authors acknowledge financial support from the following institutions:
The surveillance package is free and open-source software, and you are welcome to redistribute it under the terms of the GNU General Public License, version 2. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY.