EpiSODIC: Epidemiological Signal Observation, Detection, Identification, and Classification

[!WARNING] Under very intensive development. EpiSODIC has not been released and is not yet ready for use. Its functions, database schema, configuration and output can change substantially from one version to the next. While already fully functional, please use it only for testing until a first release is announced.

An early warning system for infectious disease outbreaks, built to run in any laboratory, in any country.

Statistical methods for spotting an unusual rise in infections are not new. The Farrington algorithm and its successors have run at national public health institutes since the early 1990s, and R has long had a solid reference implementation in the surveillance package. What has been missing is not a better method, but a system that actually runs one, day after day, in an ordinary laboratory. That signals at either ward or full-region level. That records an epidemiologist’s or surveillance scientists’s assessments. That runs reliably with clear insights for anyone involved.

EpiSODIC is that system. It takes a stream of laboratory-confirmed infections and carries it all the way from statistical detection through reconciliation into persistent clusters, structured epidemiologist assessment with a full audit trail, to an outbreak report ready for clinical colleagues, fully automated and running on a schedule of your choosing.

It Detects

Combines the Improved Farrington method with rule-based detectors for pathogens too rare for a statistical baseline, and the Moving Epidemic Method for seasonal ones such as influenza and RSV, applied automatically across every ward, institution, and region in your data.

It Reconciles

Recognises the same signal across repeated runs and across geographic levels, so one outbreak stays one dossier, not a pile of restatements.

It Assesses

Gives epidemiologists a dossier per cluster with an automatically generated interpretation of the evidence, and records every judgement made on it, kept as a full, immutable audit trail.

It Reports

Turns a confirmed cluster into an outbreak report for clinical microbiologists and infection prevention practitioners, rendered in the language your colleagues read, with every new version showing what changed since the last one. A cluster can also be put on a recurring schedule - every N days, to a list of addresses - so colleagues without a dashboard account still get their update automatically.

It Notifies

Pushes new clusters to your team the moment they arise, by email, Microsoft Teams, Slack, or even an ntfy server.

It Adapts

Works against any laboratory information system, any coding system, any administrative geography. Nothing about how it interprets your data is hardcoded.

Set-up takes minutes, and no data ever leaves your own infrastructure. The dashboard and its outbreak reports are available in English, (Modern Standard) Arabic, Dutch, French, German, Hindi, Mandarin Chinese, and Spanish.

This software is 100% free and open-source, and only allows for local storage of data and configurations. This software is perfectly safe, and all code is open to anyone willing or required to assess.

Installation

# install.packages("remotes")
remotes::install_github("certe-medical-epidemiology/EpiSODIC")

See it work, right now

No data, no credentials, no configuration - one call runs the whole system against bundled synthetic data:

EpiSODIC::episodic_demo()

This creates a temporary database, runs one detection cycle, provisions a demo epidemiologist account (printed to the console), and opens the app. Pass launch = FALSE to skip opening the app and just get a populated database path back, e.g. for scripting.

Documentation

Everything past "what is this and how do I try it" lives in the package's vignettes, so it stays organised by topic instead of one long page:

Read this...	...for
Overview	What EpiSODIC does: the Clusters and Pathogen screens' two altitudes, and the Performance, Archive, Streams, Activity and Info screens around them.
Getting your data in	The case data requirements - every column, what's optional, episodic_check_cases()/episodic_check_denominators()/episodic_check_institution_activity() to check your extract before anything runs, and episodic_add_manual_cluster() for clusters detected by another system entirely.
Deployment	Standing up a real instance: scheduling, accounts, the database backend, custom report templates.
Environment variables	The full EPISODIC_* reference table.
Detection and reconciliation	How a laboratory result becomes a dossier: the four detectors, reconciliation, and suppression.
FAQ	Answers to the questions that come up most often when getting started.

Building the package locally (or browsing offline) works the same way: vignette(package = "EpiSODIC") lists all of them, and vignette("data-format", package = "EpiSODIC") (etc.) opens one.

Licence

GPL-2.