LoginSign Up
← Back to all sparks
A

aedseo

ANALYTICS
Velocity0.0

Automated and Early Detection of Seasonal Epidemic Onset and Burden Levels

github.com/ssi-dk/aedseo ↗

An epidemic-onset detector that now brackets the whole season, not just its start

epidemiologytime-seriessurveillancer-packagegrowth-rate
◆Current state
aedseo is a Statens Serum Institut R package for automated early detection of seasonal epidemic onsets, built around growth-rate estimation over consecutive time intervals. The 1.0.0 line moved it well past its original scope: observations are now modelled as cases or population-adjusted incidence, multiple waves can be estimated in one pass, and disease-specific thresholds are computed by a dedicated function. Version 1.1.0 adds an estimate of when a season has ended after the first onset.
◆Where it's heading
The arc runs from a single-purpose onset detector toward a full seasonal-surveillance toolkit. Each release since 1.0.0 has widened what the package can say about a season rather than improving how it says it: incidence denominators, background population, multi-wave detection, thresholds, and now season end. Fixes in between are narrow and numerical, such as confidence intervals under ATLAS BLAS.
◆Prediction
The natural next step is symmetry with the onset machinery the package already has: turning the 1.1.0 seasonal offset into a first-class output alongside onset, with its own summary and plotting methods.

◆Recent moves

  1. 6mo ago

    combined_seasonal_output() now estimates when a season ends

    Adds a seasonal_offset flag to combined_seasonal_output(), tuned by burden_level_decrease and steps_with_decrease, so the package can mark the end of a season after the first onset. It extends an existing output rather than adding a new estimation path, but it is the first time aedseo reports on the back half of a season.

    View source ↗
  2. 8mo ago

    Fix growth-rate confidence intervals under ATLAS BLAS

    A single numerical fix: fit_growth_rate() computed incorrect confidence intervals when R was linked against ATLAS BLAS/LAPACK. Environment-specific and invisible to users on other builds.

    View source ↗
  3. 9mo ago

    Incidence, population adjustment and multi-wave detection land in 1.0.0

    ⚡ SPARK

    The release that set the current trajectory: to_time_series() and fit_growth_rate() take population and incidence denominators, seasonal_onset() can offset growth rates by background population, combined_seasonal_output() estimates multiple waves, and estimate_disease_threshold() is added. Every capability the later releases build on arrives here.

    View source ↗
  4. 2y ago

    Maintainership transferred to Lasse Engbo Christiansen

    An administrative handover of package maintainership with no code change. Worth noting only because the 1.0.0 expansion followed under the new maintainer.

    View source ↗
  5. 2y ago

    epi_calendar() and richer autoplot displays

    Adds epi_calendar() for classifying dates into epidemiological seasons and extends autoplot() with observed-case and growth-rate displays, alongside vignette and licensing housekeeping. Early groundwork for the season-aware framing the package later formalised.

    View source ↗
  6. 2y ago

    First CRAN release of the aeddo onset detector

    The initial CRAN release, shipping the aedseo() detector, fit_growth_rate(), the tsd time-series constructor, and predict/summary/autoplot methods. A compact single-purpose package that the 1.0.0 line would later widen considerably.

    View source ↗