Skip to contents

Cholera Epidemic Peaks Data

Source: R/data_epidemic_peaks.R
epidemic_peaks.Rd

A dataset containing identified epidemic peaks from cholera surveillance data across African countries. Peaks are detected using time series analysis with smoothing and prominence-based peak detection algorithms.

Usage

epidemic_peaks

Format

A data frame with 6 variables:

iso_code

ISO 3166-1 alpha-3 country code

peak_start

Start date of the epidemic period (Date)

peak_date

Date of peak incidence (Date)

peak_stop

End date of the epidemic period (Date)

reported_cases

Number of reported cholera cases at peak (numeric)

outbreak_interval_days

Length of the peak window in days, peak_stop - peak_start (numeric)

Source

Generated from cholera surveillance data using est_epidemic_peaks() function on combined daily surveillance data from WHO and other sources.

Details

Scope: The dataset is the full historical detection record from the surveillance time series (all years of the combined daily series, observed weeks only – AI Fourier reconstructions are excluded) and is not pre-trimmed to any particular simulation window. Consumers that score peaks against a specific config window (e.g. calc_model_likelihood(), the Python likelihood port, the simulation config builders) must filter against [date_start, date_stop] first – otherwise which.min(abs(date_seq - peak_date)) silently snaps out-of-window peaks to t=1 or t=N and biases the peak-shape likelihood terms. The internal helper MOSAIC:::.filter_epidemic_peaks() is the canonical filter and is applied at build time inside make_config_default.R, at runtime inside get_location_config(), and defensively inside calc_model_likelihood().

Epidemic peaks are identified using the following methodology:

  • Time series smoothing with 28-day running mean window

  • Local maxima detection with 10-day comparison windows

  • Prominence-based filtering (minimum 8% of maximum smoothed value for most countries; country-specific overrides below)

  • Minimum peak height threshold of 3 smoothed cases (default)

  • Minimum 75-day separation between consecutive peaks

  • Peak boundaries defined where incidence drops to 1/3 of peak height

Country-specific adjustments are applied for:

  • Niger (NER): Lower prominence threshold (1.5%) for gradual peaks

  • Cameroon (CMR): Adjusted threshold (4%) for plateau-shaped peaks

  • Ethiopia (ETH): Lower threshold (3%) for multiple outbreaks

Detected peaks sit on observed weeks: imputed (AI Fourier, assumed-zero) weeks are treated as missing (a WHO multi-week report spread over its window counts as observed), a peak day must be observed with cases > 0, and a detected peak whose window is at least half imputed days is dropped. The manual corrections below are documented outbreaks and are exempt from that window filter.

Manual corrections have been applied for known issues including:

  • Ethiopia 2024: February peak corrected to March (sustained outbreak)

  • DRC 2023: Added January peak (filtered due to proximity)

  • Nigeria 2024: Added October peak

  • Kenya 2022-2023: Added December 2022, removed June 2023 minor peak

  • Mozambique: Added February 2024 and March 2025 peaks

  • Somalia: Added major April 2017 peak and 2024-2025 peaks

  • Zambia: Added January 2018 peak

  • Tanzania: Added January 2017 peak

See also

est_epidemic_peaks for the function that generates this data

plot_epidemic_peaks for visualization

Examples

data(epidemic_peaks)
head(epidemic_peaks)
#>     iso_code peak_start  peak_date  peak_stop reported_cases
#> 1        AGO 2006-03-27 2006-04-23 2006-05-23            757
#> 174      AGO 2018-01-01 2018-01-04 2018-01-31             24
#> 2        AGO 2025-04-09 2025-04-28 2025-05-25            303
#> 3        AGO 2025-10-02 2025-10-13 2025-10-24            149
#> 4        AGO 2026-04-18 2026-04-27 2026-05-12            106
#> 5        BDI 1980-02-02 1980-02-10 1980-02-18             15
#>     outbreak_interval_days
#> 1                       57
#> 174                     30
#> 2                       46
#> 3                       22
#> 4                       24
#> 5                       16

# Countries with epidemic data
unique(epidemic_peaks$iso_code)
#>  [1] "AGO" "BDI" "BEN" "BFA" "CAF" "CIV" "CMR" "COD" "COG" "COM" "ETH" "GHA"
#> [13] "GIN" "GNB" "KEN" "LBR" "MLI" "MOZ" "MWI" "NER" "NGA" "RWA" "SDN" "SEN"
#> [25] "SLE" "SOM" "SSD" "TCD" "TGO" "TZA" "UGA" "ZAF" "ZMB" "ZWE"

# Recent peaks (2024-2025)
recent_peaks <- epidemic_peaks[epidemic_peaks$peak_date >= as.Date("2024-01-01"), ]
table(recent_peaks$iso_code)
#> 
#> AGO BDI CAF CIV CMR COD COG COM ETH GHA KEN MOZ NER NGA RWA SDN SOM SSD TCD TZA 
#>   3   4   1   1   1   3   1   1   3   1   3   3   1   3   1   4   2   3   2   3 
#> UGA ZMB ZWE 
#>   3   2   2 

# Peak duration calculation
epidemic_peaks$duration <- as.numeric(
  epidemic_peaks$peak_stop - epidemic_peaks$peak_start
)
summary(epidemic_peaks$duration)
#>    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
#>    2.00   16.00   25.00   29.34   44.00  117.00