Overview

The Outbreaks screen: one outbreak at a time

The Outbreaks screen is operational: a ranked queue of discrete outbreaks (L1 through L3), each one something a person has to make and record a decision about. That is the right unit for triage and the wrong unit for surveillance, just as a single outbreak dossier cannot tell you whether the influenza season has started.

Every outbreak opens into a dossier: case statistics, its status trajectory, an automatically generated plain-language interpretation of the evidence, the epidemic curve, similar past outbreaks and their verdicts as precedent, and the classification panel where an epidemiologist records their assessment.

The Epidemics screen: regional and seasonal

The Epidemics screen shows epidemics, the regional-scale signals at province (L4) and catchment (L5) level. An epidemic’s dossier opens with where the epidemic stands: its latest complete week against the week before, the peak so far, the MEM intensity now and at the peak, and the latest Rt. The current intensity band is also a chip in the dossier’s header (“Intensity: high”), coloured on the same five-step severity scale as an epidemiologist’s verdicts, so a band and a verdict of the same weight share a colour on every screen. Below that come the weekly case curve with the MEM thresholds drawn on it (the weeks before the epidemic began included), the epidemic’s season against earlier seasons, Rt on a log2 axis, tests and positivity, a choropleth of its cases, age and sex against the area’s own baseline, the care lines cases were found in, the contributing institutions, and the outbreaks recorded as occurring during it. Each of those outbreaks carries a “During E-123” chip on its own dossier, so the relation reads from either end.

The screen is laid out exactly like the Outbreaks screen, and an epidemic is assessed on the same form: it can be an artefact or expected variation like any other signal. Where the epidemic is seasonal, that form additionally offers the declaration - whether the season has started, has not yet started, or has ended. That is an act with policy consequences across multiple hospitals, which the system raises but never performs automatically, and it is offered only where there is a season to declare.

The Pathogen screen: one pathogen over a period

Pick a pathogen and a period, the last three months, the last twelve months, the last five years, or an exact date range, and it describes that pathogen across the whole catchment over that period:

• weekly incidence, with the Moving Epidemic Method’s pre- and post-epidemic thresholds and its medium/high/very high intensity bands drawn on it, fitted only on the seasons before the one being looked at;
• the same period laid over earlier seasons (or, for a non-seasonal pathogen, earlier calendar years) on a shared week-within-period axis, so “earlier”, “later”, “bigger”, “smaller” can be read directly;
• Rt for the pathogen as a whole rather than per outbreak, which is the population the renewal model assumes it is seeing, conditioned on case history from before the selected period;
• testing volume and positivity, age and sex against the pathogen’s own long-run distribution, geography, and care-line and institution breakdowns;
• and the outbreaks that were raised during the period, with the verdict each one received, as the way back to the operational view.

The Performance screen: is detection itself working

Detection timeliness and positive predictive value per detector and pathogen, computed from stored verdicts on outbreaks (L1-L3). Both fill in as outbreaks get assessed, so tuning EPISODIC_CONFIG against your own signal volume is an evidence-based exercise rather than a guess, once your board has classified enough outbreaks to have a baseline.

The Archive screen: last winter’s precedent

Every closed outbreak and epidemic, searchable by pathogen and place, sorted by period. A closed dossier is not filed away, it stays reachable and reasoned exactly the way an open one is, since last winter’s assessment is often the best prior evidence for judging this winter’s outbreak.

Streams and Activity: how the machinery is behaving

Streams lists every stream the lattice is currently watching (pathogen x level x institution/area/ward/region) alongside the detection configuration resolved for it at the latest run - the place to look when “why didn’t this fire” needs a concrete answer rather than a guess.

Activity is the audit log: every assessment, closure, mute, sign-in, and scheduled run, with a system action visually distinct from a person’s. Between the two, nothing the system did (or did not do) is ever only inferable from its effects.

The Info screen

The running instance’s own version, licence, and configuration, read live from the package itself - useful when reporting an issue, or simply confirming which build is deployed.

The Settings screen

The configurations set for EpiSODIC, including account management, the various notifications options, and all specific detection algorithm settings.