United Kingdom — COVID-19 weekly case surveillance

An implementation of the Farrington outbreak-detection algorithm (Farrington et al. 1996), run against real WHO/JHU-sourced case data. Every alarm below carries the full statistical reasoning behind it — observed count, expected baseline, and threshold — not a bare flag.

Omicron wave — detection detail

Oct 2021 – Feb 2022 · weekly cases

Observed Expected range Alarm

How this works

Power-transform counts (exponent 2/3) to stabilize variance.
Fit a weighted regression to the most recent 52 weeks as a rolling baseline.
Downweight past-outbreak points in a refit so old spikes don't inflate today's threshold.
Overdispersion-scaled interval at 95% significance.
Low-count guard — no alarm without real recent volume.

Full case history

2020-01-09 to present · hover any point for detail

Observed Alarm

Alarm log

Week	Observed	Expected	Upper bound	Ratio