incast

incast is an R package for infectious disease nowcasting and forecasting developed as part of Insight Net, a CDC Center for Forecasting and Outbreak Analytics initiative. It provides a unified framework for generating, evaluating, and operationalising infectious disease forecasts.

It fetches (get_data()) and validates input data (check_data()), optionally applies nowcasting to adjust for reporting delays (get_ncast()), evaluates models by cross-validation (get_cv()), and generates forecasts (get_fcast()).

Installation

You can install the development version of incast from GitHub with:

# install.packages("pak")
pak::pak("ACCIDDA/incast")

Quick start

library(incast)
tail(example_data)
#> # A tibble: 6 × 5
#>   as_of      location target          target_end_date observation
#>   <date>     <chr>    <chr>           <date>                <dbl>
#> 1 2025-12-07 CA       wk inc flu hosp 2025-12-06              233
#> 2 2025-12-14 CA       wk inc flu hosp 2025-12-06              259
#> 3 2025-12-07 NY       wk inc flu hosp 2025-12-06             1160
#> 4 2025-12-14 NY       wk inc flu hosp 2025-12-06             1171
#> 5 2025-12-14 CA       wk inc flu hosp 2025-12-13              412
#> 6 2025-12-14 NY       wk inc flu hosp 2025-12-13             1462

fcast <- example_data |>
  check_data() |>
  get_ncast() |>
  get_cv(eval_start_date = as.Date("2024-10-01")) |>
  get_fcast()
#> ℹ Using max_delay = 6 from data
#> ℹ Truncating from max_delay = 6 to 2.
#> ℹ Using max_delay = 6 from data
#> ℹ Truncating from max_delay = 6 to 2.
#> [2026-07-31 12:37:37.015] get_cv: +3.4740 secs
#> [2026-07-31 12:37:40.503] get_fcast: +4.8977 secs

fcast
#> <incast_fcast>
#> Target:   wk inc flu hosp
#> Series:   2 (location)
#> Forecast: 2025-12-20 to 2026-01-10 (h = 4)
#> Models:   3 + ENSEMBLE

fcast |> autoplot()

Save to myRespiLens format:

to_respilens(fcast, path = "respilens.json")

Citation

If you use incast in your work, please cite the package as follows:

citation("incast")
#> To cite package 'incast' in publications use:
#> 
#>   Geismar C (2026). _incast: A suite of tools for epidemic
#>   forecasting_. R package version 0.0.1,
#>   <https://github.com/ACCIDDA/incast>.
#> 
#> A BibTeX entry for LaTeX users is
#> 
#>   @Manual{,
#>     title = {incast: A suite of tools for epidemic forecasting},
#>     author = {Cyril Geismar},
#>     year = {2026},
#>     note = {R package version 0.0.1},
#>     url = {https://github.com/ACCIDDA/incast},
#>   }

Acknowledgements

The package relies on the baselinenowcast and fable framework for time series nowcasting and forecasting. It produces forecasts in the hubverse format for submission to the CDC Forecast Hubs.