Acquisition and recognition of Lycoris spp. fingerprint patterns using artificial intelligence-enhanced electrochemical sensors

Abstract

This study developed an AI-enhanced electrochemical sensing platform for the acquisition and recognition of fingerprint patterns from 17 Lycoris species. Differential pulse voltammetry (DPV) was employed to record the electrochemical fingerprints of the Lycoris pollen extracts using graphene oxide-modified screen-printed carbon electrodes. Principal component analysis (PCA) revealed distinct clustering and separation of species based on their fingerprints, with the first three PCs explaining 78.6 % of the total variance. A multilayer perceptron (MLP) neural network achieved an overall accuracy of 96.2 % in classifying the Lycoris species, while a support vector machine (SVM) model attained an accuracy of 94.8 %. The AI-based classification results showed high congruence with the traditional palynological analysis. This study demonstrates the potential of integrating electrochemical sensors with AI techniques for rapid, cost-effective, and automated plant species identification.