Project website: https://aquarom.no/
Fish farms around the world face a persistent problem: disease outbreaks that can wipe out entire production cycles in a matter of days. The conventional response has been to invest heavily in monitoring technology—more sensors, more data, more automation. But there's a catch. Despite these investments, disease-related losses haven't dropped nearly as much as they should have. In Norway, aquaculture R&D spending has increased steadily over the past decade, yet productivity growth has essentially flatlined. Something isn't adding up.
AQUAROM will tackle this problem from a different angle. Instead of assuming that better disease management requires more expensive equipment, we're going to test whether smarter use of existing, affordable technology can achieve the same results. Think of it as the difference between buying a more powerful computer and learning to write more efficient code. Both can improve performance, but one costs a lot less.
Why This Matters
The stakes are different depending on where you are. In Norway, farms operate with sophisticated sensor networks that monitor water quality, fish behavior, and environmental conditions around the clock. The technology works well, but it's expensive to install and maintain. For large operations, that's manageable. But the question is: are we using ten sensors when three would do the job just as well?
In the Philippines, the situation is almost the opposite. Small-scale tilapia and milkfish farmers can't afford elaborate monitoring systems, so they rely mostly on experience and visual observation. When disease hits, they often find out too late. The result is devastating economic losses and, in worst cases, total stock mortality. These farmers need solutions that actually fit their budgets—not simplified versions of high-end Norwegian systems, but fundamentally different approaches designed around what's realistic for them.
What We'll Do
AQUAROM will run for three years, involving partners from Norway, the Philippines, and Indonesia, working with aquaculture companies, government agencies, and farmers.
First, we'll systematically test which sensors actually matter for early disease detection and which ones are adding cost without adding value. We'll run controlled experiments comparing different sensor configurations, looking at water quality parameters like dissolved oxygen, temperature, pH, and salinity. Can minimal setups combined with smart data processing match the detection rates of more elaborate systems? That's what we'll find out.
Second, we'll develop predictive models that work even when data is incomplete. Most machine learning models assume clean, continuous data streams—fine for a Norwegian salmon farm, but not helpful for a Filipino farmer who checks water quality twice a day with a handheld meter. We'll use statistical imputation and data augmentation to build models that remain accurate despite sparse data, targeting 80% accuracy in predicting disease outbreaks.
Third, we'll study how farmers actually make decisions when using these systems. Technology only helps if people trust it. We'll conduct field trials with real farmers in both countries, observing how they interpret model outputs and what prevents adoption. This isn't just about building better algorithms—it's about building tools people will actually use.
The Philippines serves as the main testing ground for minimal-sensor approaches. Working with BFAR and local fish farms, we'll deploy prototype systems and track performance over multiple production cycles. Then we'll reverse-test the approach in Norway using real data from salmon farms. If a farm currently has twenty sensors, can we downsample to eight and maintain equivalent disease detection rates? If so, that's evidence the principle works across very different contexts.
What We Expect to Learn
If this works, several things will become clear. Small-scale farmers in developing countries will gain access to disease monitoring that's actually affordable—not next year's dream technology, but systems they can implement now. Norwegian farms will discover they can probably cut monitoring costs by 30-50% without sacrificing performance. And the broader aquaculture industry will have a new framework for thinking about technology: effectiveness per dollar spent, not just raw capability.
The project will produce practical tools: open-source software for disease prediction that works with minimal data, guidelines for sensor placement that optimize detection while minimizing cost, and detailed cost-benefit analyses. Three PhD students will complete dissertations through this project, focusing on sensor optimization, epidemiological modeling, and predictive analytics.
The Team
NORCE Norwegian Research Centre leads the project, bringing expertise in numerical modeling, sensor systems, and data analytics. Partners include the University of Southern Philippines Foundation, Iloilo State University of Fisheries Science and Technology, and Universitas Brawijaya in Indonesia.
On the industry side, we're working with Manolin AS (a Norwegian aquaculture analytics company), BFAR Region 7, and HOC PO Feed Corp in the Philippines. This combination of academic research, government support, and industry partnership should let us move quickly from concept to implementation.
