TellSpec's AI sensor detects fish fraud in real-time: 'This technology is disruptive,' says CEO
“There are four main types of fish fraud that occur in the supply chain,” Isabel Hoffmann, CEO of UK-headquartered TellSpec, told FoodNavigator at the EU’s European Institute of Innovation and Technology (EIT) INNOVEIT awards in Budapest last week, where she won first prize in the EIT Woman category.
“We are working to address the first three types of fraud right now."
Number one on the list of ways to adulterate fish is by injecting it with water before it is frozen. This can add up to 50% to its weight and frozen fish that contains more than 10% is considered to have been adulterated.
The second is species substitution whereby fish is intentionally mislabelled. It is estimated that one-third of the fish sold worldwide is intentionally mislabelled. In Europe, the most common form of species substitution occurs with the catfish, pangasius.
“Europe is the largest importer of pangasius in the world," Hoffmann said. “In 2017, we imported 245 million euros-worth of pangasius but it is sold as eighteen different species of fish, from cod to sole to grouper.”
The third form is the misuse of antibiotics, where fish are farmed in poor conditions and given antibiotics, some of which are not allowed in Europe, but then sold as wild.
TellSpec’s patent-protected tools combine Near Infrared Spectroscopy (NIR) spectroscopy, bioinformatic techniques and artificial intelligence-based algorithms to analyse foods at the molecular level in real time.
After selecting the relevant parameters for the sample – species, fresh or frozen, whole or fillet, with skin or skinless – the user places the NIR spectrometer on the fish sample and pushes a button to begin the analysis.
“It sends light at a consistent energy, which is partly absorbed by the molecules of the fish and reflected back," Hoffmann explained while giving a demonstration. "This information is sent [via Bluetooth] to complex, statistical machine-learning models in the cloud that do the analysis by counting the photons, […] which gives us a fingerprint. Our machines can predict in an intelligent way what that fingerprint represents in terms of moisture, excess water, protein, etc, and then send it back to the app.”
The user's mobile phone immediately presents the data in the form of graphs and tables.
'This will be big'
TellSpec predicts significant demand for its sensor, initially in Europe and eventually China, the world’s biggest fish market.
“This will be big. Fish fraud and adulteration is the second biggest form of food fraud in the world after olive oil,” she said. “Our technology is very disruptive since currently there are no rapid, portable affordable sensors that the fish supply chain can use.”
Users can buy different ‘kits’ that analyse the sample for excess water content, decay, antibiotics and can even detect if a fish is pangasius (although this last kit destroys the fish being tested and so could be used to test one sample in a batch).
The sensor costs €1,300 and its accompanying app is free to download with users paying between €0.09 and €1.20 per analysis.
“The idea is we will develop many kits, one for pangasius, one for cod and so on."
The tool also gives a breakdown of the sample’s nutrient content. “The fat content could probably tell us if a fish is farmed or wild. Farmed fish has a higher fat content as they are fed much faster.”
TellSpec already has a number of products on the market, including tools that measure the sweetness or acidity of fruit and vegetables, detect the presence of melamine in infant formula and flour (the chemical responsible for the deaths of six infants in China in 2008) or urea and hair in flour.
Its fish fraud product, which was funded by EIT Food and has been developed in partnership with Spanish Azti, Italian Microbion, UK retailer Waitrose and the Institute for Global Food Security (IGFS) of Queen’s University of Belfast, however, is scheduled to launch mid-2019.
“Fish is far more complex than fruit, vegetables, cheese and nuts. There is frozen, defrosted and fresh fish, whole fish and fish in fillets, skin and no skin... and the list goes on. For each category and species, we had to do a model. So the data collections and testing was the intensive part of the project.”
According to Hoffmann, the fourth biggest form of fish fraud relates to illegal, unreported fishing where fish and seafood are caught in protected areas by unlicensed boats. "We will address this one next year with blockchain," she said.