Spain's new algorithm names an invading mosquito from a phone photo in five minutes

By David Ogilvy, Chief Marketing Officer at Mosticare Global | Published 2026-06-02

A person in Spain photographs a mosquito on their kitchen wall, opens an app, and uploads the picture. Four minutes later an algorithm has looked at it. Within five, the photographer has an answer — and if the mosquito is an invasive species turning up somewhere it has not been recorded before, a regional health department has an alert. No entomologist had to be awake. This is AIMA, the artificial-intelligence engine Spain's Ministry of Health has just folded into its national citizen-science surveillance network, and it is one of the more quietly impressive things to happen in European vector control this year.

The announcement came from the Ministerio de Sanidad and the Mosquito Alert project. It arrived alongside two facts that explain why Spain felt the need to build it: the tiger mosquito, Aedes albopictus, has now been confirmed in 156 municipalities since 2023, and a second invasive species, the Japanese mosquito Aedes japonicus, has been detected in 10 municipalities across Asturias, Cantabria and the Basque Country — the cool, green north, not the warm Mediterranean coast where people expect this story to unfold.

For Mosticare's readers, the interesting thing is not that Spain has more mosquitoes. It is how the country found out — and what that machinery says about where surveillance is heading.

What AIMA does, and how well

Mosquito Alert is not new. For years it has run on the same principle as a birdwatching app: ordinary people photograph mosquitoes, upload them, and a panel of expert entomologists confirms the species. The data feed a public map and a national early-warning system. The weakness was always the bottleneck. Every photograph needed a human expert, and humans are finite. As the uploads climbed into the tens of thousands, the queue grew.

AIMA — an AI for the identification and monitoring of mosquitoes — is built to break that bottleneck. The system processes incoming observations every four minutes and returns a verdict to the user in under five. It has been trained to recognise the species that matter: Aedes albopictus, the Culex genus that carries West Nile, the newly arrived Aedes japonicus — which it learns to tell apart from its near-twin Aedes koreicus — and even Aedes aegypti, the yellow-fever mosquito, which is not yet established in Spain but which the country plainly wants its detector primed for before it arrives.

Now the honest performance figures, because they are more revealing than a rounded-up "AI accuracy" boast. Of every 100 tiger-mosquito images AIMA receives, it classifies 55 correctly. Of those, the 25 it is most confident about — above 98% certainty — are published straight to the public map with no human in the loop. The remaining cases are escalated to the expert entomologists who ran the system before. In other words, AIMA is not replacing the specialists. It is clearing the easy, high-confidence quarter off their desks so they can spend their attention on the genuinely ambiguous photographs. That is a sober, well-designed division of labour — and a more trustworthy claim than "the machine does it all."

The scale the machine is feeding on

The reason the bottleneck mattered is in the participation numbers. Through the Mosquito Alert app, Spanish citizens have logged more than 33,600 observations: over 19,000 in 2023, nearly 10,000 in 2024, and more than 4,600 in the first months of 2025. Each one is a data point an entomologist would otherwise have had to chase in the field. The whole apparatus is, in effect, a national sensor network made of phones and curiosity, with an algorithm now sitting at the front desk.

This is worth pausing on, because it inverts a tired assumption. The headlines about mosquitoes and technology usually involve gene drives, sterile males, or Wolbachia releases — clever interventions that do something to the mosquito. AIMA does nothing to the mosquito at all. It simply sees it faster. And seeing it faster is, unglamorously, the thing that decides whether a newly arrived species is caught in its first season or its third. Surveillance is the least cinematic part of vector control and very often the most decisive.

The Mosticare lens: knowing is half of stopping

Here is where the story touches ordinary readers. Aedes japonicus establishing in Cantabria is not a Mediterranean problem. It is a northern, temperate, cool-climate problem — which is to say, a preview of exactly the conditions much of central and northern Europe shares. The species is a competent vector under laboratory conditions and tolerates cold better than its tropical cousins. Spain's value to the rest of Europe right now is not its weather. It is its method: a working, public, AI-assisted citizen network that turns a phone photograph into a surveillance signal in five minutes.

And the practical lesson for a household is encouraging, because it asks so little. The same standing water that breeds these mosquitoes — the saucer under the flowerpot, the blocked gutter, the bucket left in the rain — is also the thing an individual can remove in an afternoon. Surveillance maps the threat; source reduction and physical barriers remove it. Spain's system is good at the first half. The second half has always been, and remains, the part that happens at home.

There is a temptation, watching the map fill in, to read it as defeat — another front lost, another species arrived. The better reading is the opposite. A mosquito that is photographed, classified and mapped within five minutes is a mosquito that is no longer spreading in the dark. You cannot manage what you cannot see, and Spain has just made itself able to see a great deal more, a great deal faster.

What to watch next

Two things. First, whether AIMA's published accuracy improves as the training set grows — 55-in-100 on tiger mosquitoes is a respectable start for a system designed to defer its uncertain cases, but the number to watch is how the high-confidence, auto-published share climbs over the coming season. Second, and more importantly for the rest of the continent: whether other European countries adopt the model rather than rebuilding it. Mosquito Alert already has sister projects beyond Spain. An AI-assisted citizen network is the rare piece of public-health infrastructure that gets cheaper and better the more places run it, because every new photograph anywhere improves the detector everywhere.

The Ministry framed the launch as Spain consolidating itself "as a European reference in this field." On the evidence of AIMA, that is not a boast. It is a fair description of a country that has built something the rest of Europe should copy.

A note on sourcing: this piece draws on Spanish-language announcements from the Ministerio de Sanidad and iSanidad. The English here is our own; Babel may seed translated versions.

Sources cited

1. Ministerio de Sanidad, nota de prensa on AIMA and invasive mosquito surveillance (Spanish), May 2026 — https://www.sanidad.gob.es/gabinete/notasPrensa.do?id=6704&metodo=detalle
2. iSanidad, El mosquito tigre se expande hasta 156 municipios en España y el mosquito de Japón se detecta en Asturias, Cantabria y País Vasco (Spanish), May 2026 — https://isanidad.com/335913/el-mosquito-tigre-se-expande-hasta-156-municipios-en-espana-y-el-mosquito-de-japon-se-detecta-en-asturias-cantabria-y-pais-vasco/