A team of researchers from Banaras Hindu University and Sikkim University developed a new, cost-effective, and home-based method for detecting adulterants in milk using the evaporation process and machine learning algorithm.
The method has been developed by Tapan Parsian a research scholar under the supervision of Archana Tiwari of the department of Physics, in collaboration with Ajay Tripathi, of Sikkim University.
They claimed that the method is useful in ensuring the strength and quality of milk and public health.
Tiwari said that the evaporation process is a phenomenon in which, when a drop is left to evaporate on a surface, the suspended particles in the drop move towards the edges and leave behind a ring-like pattern on the surface. Different types of suspended particles form their own unique evaporated ring patterns. These evaporated ring patterns act as fingerprints for identifying the substance.
She said that to validate the method, synthetic milk was prepared in the lab by mixing appropriate amounts of detergent, ordinary vegetable oil, urea, and tap water to create a mixture that resembled real milk. This synthetic milk was then mixed with real milk in varying quantities. The evaporated ring patterns of the synthetic milk were then analyzed. During the analysis, several signs were identified in the evaporated ring patterns that reveal the quality of the milk.
She said that the synthetic milk evaporation patterns exhibit multiple rings, visible micro-droplets of vegetable oil, and varying degrees of transparency, which can provide insights into the extent of adulteration in pure milk. If the evaporated patterns are highly transparent, it indicates a higher proportion of synthetic milk adulteration. Conversely, if the patterns are less transparent, it suggests a lower presence of synthetic milk. This method provides a simple home-based approach for milk quality assessment.
During the experiment, for large-scale analysis, the water and synthetic milk mixed evaporated ring patterns were used to train and validate the machine learning model.