Eating spoiled beef is dangerous, but there are currently no simple and effective methods for assessing the freshness of beef. Scientists from the Gwangju Institute of Science and Technology (GIST), South Korea, have solved the problem by developing a unique AI (artificial intelligence).
Although beef is one of the most popular foods in the world, it can be dangerous. Improper storage and eating of stale tenderloin can lead to serious health problems. Unfortunately, there are many drawbacks to the available methods for testing beef freshness. For example, chemical analysis or assessment of the microbial population is too time-consuming and requires professional skills. On the other hand, non-destructive beef methods based on near infrared spectroscopy require expensive and sophisticated equipment. Could artificial intelligence be the key to a more cost-effective way to measure the freshness of beef?
In South Korea, a team of scientists has developed a new strategy that combines deep learning with diffuse reflection spectroscopy (DRS), a relatively inexpensive optical technique. Unlike other types of spectroscopy, DRS does not require complex calibration. Instead, it can be used to quantify a fraction of the molecular composition of a sample using only an affordable and easily customizable spectrometer. Details of the new method are published by Food Chemistry.
To determine the freshness of beef samples, the scientists used DRS measurements to estimate the proportion of different forms of myoglobin in meat. Myoglobin and its derivatives are proteins that are mainly responsible for the color of meat and its changes during decomposition. However, manually converting DRS measurements to myoglobin concentration for final determination of sample freshness is not a very accurate strategy, and this is where deep learning comes into play.
Convolutional neural networks (CNNs) are widely used artificial intelligence algorithms that learn from a pre-classified dataset – a training dataset – and find hidden patterns in the data to classify new inputs. To educate CNN, the researchers collected data on 78 beef samples in the spoilage process by regularly measuring their pH (acidity) along with DRS profiles. They combined the findings with estimates of myoglobin. As a result, the deep learning algorithm correctly classifies the freshness of beef samples in seconds in about 92% of the time.