AI-powered Verification System for Medicinal and Edible Homologous Substances: Evidence-based Identification, Data Standardization and Web-based Practice

The rapid expansion of the health preservation market has been accompanied by widespread misinformation, vague efficacy claims, insufficient evidence support, and inadequate standardization in the medicinal and edible homologous (MEH) industry. To address these challenges, this study developed an AI - assisted verification system that integrates multi - source data standardization, deep learning - based identification, and lightweight Web application deployment.A unified data specification covering 100 representative MEH materials was established by synthesizing traditional Chinese medicine (TCM) classics, modern clinical studies, and pharmacopoeial standards, raising the data standardization rate from 62% to 95%.A hybrid deep learning model incorporating a convolutional neural network and attention mechanism was constructed for component recognition, efficacy matching, and contraindication warning. Independent testing showed that the model achieved 91.0% accuracy in component identification [1], 93.0% accuracy in efficacy matching, and 93.0% accuracy in contraindication warning.A lightweight Web platform was further developed to enable real - time query, evidence - level annotation, and science popularization. User evaluation indicated that 87.0% of respondents recognized the practicability and reliability of the platform.This study establishes a systematic and replicable framework of "data standardization–model validation–platform deployment," which provides a feasible technical route for the scientific verification of MEH materials and the governance of pseudoscientific health claims. The findings contribute to enhancing transparency and standardization in the MEH industry and offer a scalable, lightweight innovation paradigm for digital TCM and health informatics research.