Package: EpiSODIC 0.24.0

Checking what they actually delivered

Before you change anything | Too many false alarms | The Farrington bound is too narrow | same_place fires on endemic patterns | Suppression lets redundant signals through | Missing outbreaks | Farrington never fits | same_place thresholds are too high | The eligibility gate excludes a stream | Clusters close too early or too late | Clusters close before the outbreak is really over | Clusters stay open long after the last case | Cooldown absorption swallows a real recurrence | Priority ranking does not match your triage | Seasonal detection is wrong | MEM never runs | MEM fires too early or too late | Per-pathogen parameters | Verifying that a change helped | Against synthetic ground truth | Against your own board's verdicts | Reference | Reconciliation | Eligibility | Effect-size floor | same_place detector | Farrington detector | MEM detector | rare_trigger detector | Nowcast | Outbreak-end probability | Epidemic direction | Epidemic outlook | Typical past season | Priority score | Scale | Geography | Suppression | Notifications | Report | Access | Database | Per-pathogen CSV columns | See also

The lattice: streams, not one big feed | Four detectors, not one algorithm | Scale routing: outbreaks and epidemics | Reconciliation: turning detections into outbreaks and epidemics | Epidemic closure | The "during" link | Suppression: one lattice, not five parallel views | Derived state, not a stored status | Declarations: a human act

No dependency on any one laboratory system | Three altitudes | The Outbreaks screen: one outbreak at a time | The Epidemics screen: regional and seasonal | The Pathogen screen: one pathogen over a period | Beyond the three altitudes | The Performance screen: is detection itself working | The Archive screen: last winter's precedent | Streams and Activity: how the machinery is behaving | The Info screen | The Settings screen | Where to go next

The two things you provide | Configuration lives outside the repository | Accounts | Closing the app to anonymous visitors | Watching sign-ins | Where the database lives | Which servers this is tested against | Sharing a schema with something else | If you already have a MariaDB or MySQL database | Upgrading EpiSODIC | What a migration does, and does not do | When the schema is not the application's to change | Going back | Running the app | Behind a reverse proxy | Custom report templates | Scheduled reports | Optional pieces, and their fallbacks | Notifications | See also

Getting your data in | Running it day to day | Accounts and hosting

Cases (mandatory) | source_key: one identifier per result, not per specimen | One culture, more than one reported result | Deduplication is EpiSODIC's job, not yours | Check your data first | Positivity metadata (optional) | Institution activity (optional) | Geographic reference data (optional) | The choropleth map | L4 (province-level) detection | What a run reports | Clusters from another system (optional) | See also

Where to configure | Minimal example: ntfy | ntfy with authentication | SMTP (Gmail, institutional relay, any SMTP server) | sendmail (Linux/Unix) | Microsoft 365 | Option A: App-only (client credentials) -- recommended for unattended cron | Option B: Delegated (cached interactive token) | Option C: Reuse an already-cached login (no app registration needed) | Microsoft Teams | Setting up the webhook | Slack | Triggers | Dashboard link | Full reference example | Testing your setup | What a notification looks like | New clusters | Run failure | Colours and font | Your own email layout | Troubleshooting | See also

Setting up a schedule | Cadence and delivery | What changed since last time | The email itself | Custom report templates and patient-level detail | See also

What it does | The ground truth is data | The two matching rules | What it reports, and how to read it | A comparator, and a control | Which detector earns its place | Where the real numbers live