Sirish Subash Wins 2024 Young Scientist Challenge

According to the 3M and Discovery Education announcement, Pestiscand uses spectrophotometry to measure light at different wavelengths reflected from the surface of fruits and vegetables. A machine learning model analyzes the resulting spectral data to determine whether pesticide residue is present. The handheld prototype includes a sensor, power supply, display, and processor.

3M reported that testing on spinach and tomatoes achieved pesticide-residue identification accuracy greater than 85%. Spectroscopy Online similarly described the system as a non-destructive spectrophotometric approach paired with ML analysis of spectral data.

The reported result is limited to the tested produce types and conditions described in the challenge materials. In applied ML and analytical sensing, performance across two produce categories does not by itself establish robustness across pesticide compounds, residue concentrations, cultivars, surface textures, lighting conditions, or field samples. Independent validation, including sensitivity, specificity, calibration procedures, and comparison with laboratory reference methods, would be required to assess broader practical performance.

Pesticide-residue testing is commonly associated with laboratory analytical workflows. Pestiscand's project framing centers on a portable, non-destructive alternative for produce screening. The 3M announcement described the device as designed to detect residues without damaging the item being tested.

For ML practitioners, the project is an accessible example of a familiar edge-AI pattern: acquire a physical signal, transform it into model inputs, and return a classification result on a compact device. The central technical challenge in comparable systems is not only model accuracy, but also whether the training and evaluation data represent the variability encountered outside controlled tests.