Estimate Epidemic Peaks from Cholera Time Series Data
Source:R/est_epidemic_peaks.R
est_epidemic_peaks.RdThis function identifies and characterizes epidemic peaks in cholera incidence time series data for all countries with available data. It detects peak timing, magnitude, and duration of epidemic periods.
Details
Peaks are detected on observed weeks only: imputed days – those whose
disaggregation_method is a modelled method such as an AI
fourier_* reconstruction or assumed_zero – are treated as
missing, and a detected peak whose [peak_start, peak_stop] window is at
least half imputed days is dropped. A WHO multi-week report spread over the
weeks it covers (who_catchup_*, see process_WHO_weekly_data)
counts as observed: its total is a reported count over a known window, whereas
blanking it would carve a false trough into the outbreak it belongs to. An even
spread creates no local maximum of its own; a shaped spread follows another
source's weekly counts or a documented epidemic curve, so its maxima are real.
A detected peak day is the centre of any flat stretch of the smoothed curve and
must itself be observed with cases > 0. The hand-curated peaks the function
appends (documented outbreaks) are exempt from the imputed-window filter.
Examples
if (FALSE) { # \dontrun{
# Set up paths and run epidemic peak estimation
set_root_directory("~/MOSAIC")
PATHS <- get_paths()
# Estimate epidemic peaks for all countries
epidemic_results <- est_epidemic_peaks(PATHS)
} # }