The last mile of epidemic intelligence

Over the past decade, the global health community has made real progress on detection. Verified outbreak reporting is faster, laboratory networks are better connected, and open source signals surface events that would once have stayed local for weeks. Detection is no longer the scarce resource.

What remains scarce is the step immediately after detection. A district health officer who receives a confirmed signal at 8am still has to decide, that same day, where to send a limited team, which of four competing hypotheses to test first, and which activity to stop in order to free the staff. Very little of the intelligence produced upstream is shaped to answer those questions.

Where the distance opens up

In our field work the gap is consistently structural rather than technical. Four patterns recur:

• Signals arrive without operational context. A confirmed case count says nothing about which neighbourhoods are reachable this week, or which facility has a functioning cold chain.
• Laboratory results and field observations live in separate systems, and are reconciled by memory rather than by method.
• Confidence is implied rather than stated, so decision makers cannot tell a strong finding from a working assumption.
• Nobody owns the question of what to stop doing, which is usually the decision that actually frees capacity.

Treating the decision as the deliverable

Applied epidemic intelligence starts from the opposite end. Instead of asking what can be measured, it asks what decision is imminent, who owns it, and what evidence would change it. Everything upstream, the signal intake, the field validation, the multimodal capture, is then instrumented to serve that decision rather than to complete a dataset.

The measure of intelligence is not how much it explains. It is whether the next decision changed because of it.

This is why our work is embedded rather than remote. Field validation is not a quality control step performed on someone else's data, it is the point at which a signal acquires the operational texture that makes it actionable: access, capacity, trust, seasonality, and the specific vulnerabilities of the population in front of you.

What this implies for institutions

For governments and public health institutes, the practical implication is that additional surveillance investment yields diminishing returns unless matched by capacity to prioritize. For humanitarian and global health organizations, it means deployment decisions can be defended on evidence rather than on precedent. And for laboratories, it means diagnostic output has to be designed backwards from the operational questions it will be asked to settle.

None of this requires new detection technology. It requires closing the last mile.