In cmu-delphi/epiprocess: Tools for basic signal processing in epidemiology

source(file.path("vignettes", "_common.R"))
knitr::opts_chunk$set(
  fig.path = "man/figures/README-"
)

epiprocess

The {epiprocess} package works with epidemiological time series data and provides tools to manage, analyze, and process the data in preparation for modeling. It is designed to work in tandem with epipredict, which provides pre-built epiforecasting models and as well as tools to build custom models. Both packages are designed to lower the barrier to entry and implementation cost for epidemiological time series analysis and forecasting.

{epiprocess} contains:

• epi_df() and epi_archive(), two data frame classes (that work like a {tibble} with {dplyr} verbs) for working with epidemiological time series data
• epi_df is for working with a snapshot of data at a single point in time
• epi_archive is for working with histories of data that changes over time
• one of the most common uses of epi_archive is for accurate backtesting of forecasting models, see vignette("backtesting", package="epipredict")
• signal processing tools building on these data structures such as
• epi_slide() for sliding window operations (aids with feature creation)
• epix_slide() for sliding window operations on archives (aids with backtesting)
• growth_rate() for computing growth rates
• detect_outlr() for outlier detection
• epi_cor() for computing correlations

If you are new to this set of tools, you may be interested learning through a book format: Introduction to Epidemiological Forecasting.

You may also be interested in:

• {epidatr}, for accessing wide range of epidemiological data sets, including COVID-19 data, flu data, and more.
• rtestim, a package for estimating the time-varying reproduction number of an epidemic.

This package is provided by the Delphi group at Carnegie Mellon University.

Installation

To install:

# Stable version
pak::pkg_install("cmu-delphi/epiprocess@main")

# Dev version
pak::pkg_install("cmu-delphi/epiprocess@dev")

The package is not yet on CRAN.

Usage

Once epiprocess and epidatr are installed, you can use the following code to get started:

library(epiprocess)
library(epidatr)
library(dplyr)
library(magrittr)

Get COVID-19 confirmed cumulative case data from JHU CSSE for California, Florida, New York, and Texas, from March 1, 2020 to January 31, 2022

df <- pub_covidcast(
  source = "jhu-csse",
  signals = "confirmed_cumulative_num",
  geo_type = "state",
  time_type = "day",
  geo_values = "ca,fl,ny,tx",
  time_values = epirange(20200301, 20220131),
  as_of = as.Date("2024-01-01")
) %>%
  select(geo_value, time_value, cases_cumulative = value)
df

Convert the data to an epi_df object and sort by geo_value and time_value. You can work with an epi_df like you can with a {tibble} by using {dplyr} verbs

edf <- df %>%
  as_epi_df(as_of = as.Date("2024-01-01")) %>%
  arrange_canonical() %>%
  group_by(geo_value) %>%
  mutate(cases_daily = cases_cumulative - lag(cases_cumulative, default = 0))
edf

Compute the 7 day moving average of the confirmed daily cases for each geo_value

edf <- edf %>%
  group_by(geo_value) %>%
  epi_slide_mean(cases_daily, .window_size = 7, na.rm = TRUE) %>%
  rename(smoothed_cases_daily = slide_value_cases_daily)
edf

Autoplot the confirmed daily cases for each geo_value

edf %>%
  autoplot(smoothed_cases_daily)

cmu-delphi/epiprocess documentation built on Oct. 29, 2024, 5:37 p.m.