COF-AgNPs Nanoenzymes Combined with Multimodal Machine Learning for Dual-model Detection of Hg 2+ in Aquatic Products

Imagine you are a detective trying to solve a mystery in a bustling, chaotic kitchen. The culprit is a silent, invisible poison that loves to hide in our seafood, making its way from the ocean to our dinner plates. This poison is mercury, a sneaky element that can damage our brains and bodies even in tiny amounts. For years, catching this culprit required sending samples to a high-tech lab, using giant, expensive machines that sound like they belong in a sci-fi movie. It was slow, costly, and needed a team of experts just to read the results. But recently, scientists have started looking for a smarter way: using tiny, artificial "mini-machines" called nanozymes. Think of these as microscopic robots that act like natural enzymes (the body's own chemical workers) but are built from man-made materials. They are cheap, sturdy, and can be programmed to change color when they spot a specific bad guy. To make this even better, scientists are now teaching computers to act as super-smart assistants, using "machine learning" to look at the data and spot patterns that human eyes might miss, especially when the "kitchen" (the seafood sample) is full of other messy ingredients that could confuse the test.

In this study, a team of researchers from Jiangsu Ocean University built a brand-new detective tool to catch mercury in fish and shellfish. They created a special hybrid material called COF-AgNPs. You can picture this as a microscopic, sea-urchin-shaped sponge (the COF) that is covered in tiny silver nanoparticles (AgNPs). When this sponge meets mercury ions (Hg²⁺), something magical happens: the silver and mercury lock together to form a silver-mercury alloy. This alloy acts like a turbocharger for the sponge, supercharging its ability to act like an enzyme. In a simple test, this super-charged sponge turns a clear liquid blue by reacting with a substance called TMB. The more mercury there is, the bluer the liquid gets. The team found that this reaction happens incredibly fast, in just 5 minutes, and is so sensitive it can detect mercury at a level as low as 10.3 nM. They also proved that this little sponge is very picky; it ignores other common metals like copper or zinc, only reacting to mercury, which is crucial because real seafood is full of other minerals that usually mess up tests.

But the researchers didn't stop at just making a color-changing sponge. They knew that real-world seafood is a messy mix of proteins, fats, and minerals that could confuse a simple color test. So, they teamed up with a computer brain. They took two types of data from their test: the exact shade of blue measured by a machine (absorbance) and the color captured by a smartphone camera (RGB values). They fed this "dual-model" data into a machine learning algorithm. It's like giving the computer two different pairs of glasses to look at the same problem, helping it see through the confusion of the seafood matrix. The computer learned to recognize the specific "fingerprint" of mercury, filtering out the noise. The result was a Support Vector Machine (SVM) model that could predict the mercury levels with an accuracy of 96.08%.

The team tested their new system on five different types of real aquatic products: shrimp, squid, yellow catfish, clams, and kelp. They added known amounts of mercury to these samples to see if their tool could find it. The results were impressive: the system could detect mercury across a wide range, from 0.1 to 50 µM, with recovery rates between 90.18% and 103.22%. This means the tool is not only fast and cheap but also reliable enough to be used in the real world. By combining a smart, color-changing nano-sponge with a computer that learns from data, the researchers have created a powerful new way to keep our seafood safe, moving away from slow, expensive lab tests toward a future where we can quickly and accurately spot dangerous pollutants right where they are found.