Why Disease Intelligence Must Connect to Field Action

Public health has never had more information about disease risk. Digital surveillance platforms aggregate reports from thousands of facilities. Satellites measure rainfall and land surface temperature daily. Open-source intelligence tools scan news and social media in dozens of languages. Forecasting models are published every season.

And yet, in programme after programme, the same pattern repeats: the surge is recognized, the report is written, the meeting is held - and the response arrives after the peak.

The problem is rarely a lack of intelligence. It is the distance between intelligence and action.

Where the gap opens

Consider a typical sequence in a vector-borne disease programme. A district surveillance officer notices positivity rising in weekly reports. A state epidemiologist confirms it a week later when data is aggregated. A forecast, if one exists, sits in a slide deck. A decision to intensify vector control is made at a coordination meeting. Then someone has to work out which wards, which teams, how many commodities and in what order - usually in a spreadsheet, often from scratch.

Each hand-off introduces delay and loses information. The forecast that explained why risk was rising, and where it was concentrated, rarely survives the journey to the microplan. And once field teams are deployed, the programme often has little visibility of what was actually done until reports arrive weeks later.

Intelligence that stops at the dashboard

Many tools in disease intelligence are designed to inform: maps, alerts, briefings. That is valuable - but informing is not the same as enabling. A risk map does not tell a supervisor which team to send to which settlement tomorrow. A forecast does not reconcile itself with the stock levels in a district store.

When intelligence stops at the dashboard, three things happen:

• Decisions default to habit. Without a practical path from risk to plan, programmes fall back on last year’s allocation.
• Accountability blurs. It becomes hard to say whether a poor outcome reflected bad intelligence, a poor decision or weak execution.
• Learning stalls. If activities are not linked to the risk that motivated them, there is no way to learn which responses worked.

What connected intelligence looks like

Closing the gap does not require a single monolithic system. It requires that a few things are deliberately connected:

1. Verified signals feed forecasts. Events confirmed by analysts should update risk estimates, not sit in a separate inbox.
2. Forecasts feed prioritization. Risk should be combined with vulnerability and access to care using transparent, agreed criteria.
3. Priorities feed microplans. Ranked areas should translate into draft field plans that account for teams, distance and stock.
4. Execution feeds monitoring. Planned versus completed activities should be visible while there is still time to adjust.
5. Outcomes feed the next cycle. Coverage, cost and epidemiological trends should inform the next round of planning.

None of these links is technologically exotic. What has been missing is a design philosophy that treats the whole chain as the product.

Keeping humans in charge

Connecting intelligence to action raises a fair concern: does automation remove judgement from decisions that affect communities? It should not. The aim is to remove friction, not people. Draft plans should be reviewed by supervisors who know the terrain. Rankings should expose their weights so that stakeholders can challenge them. Forecasts should show their uncertainty so that decision-makers can weigh them appropriately.

Done well, a connected system gives experienced people more time for the judgement only they can exercise - and a clearer record of why decisions were made.

The measure of success

The value of disease intelligence is not the number of alerts produced or the sophistication of a model. It is whether the right intervention reached the right community at the right time - and whether we can show it.

That is the standard Epinova is being built to meet. We are starting with malaria and dengue because both are strongly shaped by local conditions and both reward precise, timely action. Over the coming months we will be working with partners to test whether connecting each step in the chain makes a measurable difference - and we will report what we learn, including what does not work.