Enhanced Biomimetic Sensor for Cigarette Brand Differentiation using CdSe Quantum Dots and Machine Learning

Abstract

Abstract In our doctoral research, we ex- plored the combined influence of CdSe Quantum Dots (QDs) and the reactions of organic substances to create an olfactory-inspired sensing system based on the mammalian olfactory framework. Our primary aim was to identify the complex composition of volatile organic com- pounds (VOCs) found in cigarette smoke. We designed an innovative optical olfaction device to categorize various VOCs emitted by cigarettes. We utilized advanced techniques like Unsupervised Independent Compo- nent Analysis (ICA) and supervised Linear Discriminant Analysis (LDA) for robust data analysis. The LDA yielded remarkable results, with 100% precision in both the training and cross-validation phases. To validate our system, we rigorously assessed its ability to distinguish between five different cigarette brands, achieving 100% precision in training and an im- pressive 85% during cross-validation. Using LDA, we also conducted a comprehensive analysis of 100 sam- ples of four popular Indian cigarette brands (Gold Fake, Four Square, Navy Cut, ITC Classic), including authentic and counterfeit variants, resulting in a commendable 97% accuracy. Our analytical protocol is efficient, cost-effective, user-friendly, and highly reliable. The remarkable selectivity of our sensor array makes it indispensable for detecting genuine and counterfeit cigarettes, providing crucial support for global border control efforts.