Machine learning-assisted visual sensor array for identifying the origin of Lilium bulbs

In this work, we constructed a machine learning-assisted dual-channel visual sensor array for identifying the origin of Lilium bulbs (BH). Nanogold clusters (AuNCs) and quantum dots (QDs) were selected and combined into sensor arrays. The amino acids existed in lilium bulbs could induce aggregation-induced fluorescence enhancement effect (AIEE) of AuNCs through hydrogen bonding. The hydrogen protons released from the amino acids and phenolic acids could interact with the COO- group of QDs, resulting in aggregation-caused quenching (ACQ) of QDs. Due to the different contents of amino acids and phenolic acids in BH from different origins, the sensor array can produce distinct and different fluorescent colors, such as red, blue, green and purple. In conjunction with pattern recognition by the RF model, the sensor array clearly identifies the origin of BH with 94.4% prediction accuracy. The visual sensor array constructed in this work exhibited the advantages of simplicity, speed, accuracy and portability, showing potential application prospects in identifying the origin of food and traditional Chinese medicine.