化学计量学
模式识别(心理学)
人工智能
融合
传感器融合
质量(理念)
化学
信息融合
计算机科学
数学
色谱法
生物
质量评定
高光谱成像
皮质(解剖学)
作者
Pan-pan Wang,Xuehua Fan,Jianing Sun,LI Meng-Han,Xin-jing Gui,Yong-qin Zhu,Qing-xiao Wang,Xiao-yao Wang,Rui-xin Liu
标识
DOI:10.1016/j.jfca.2026.109692
摘要
Cinnamomi Cortex (CC ) is a widely used medicinal and edible species with substantial market potential and frequent clinical use. Therefore, the quality consistency of CC formula granules (CCFG) directly affects therapeutic efficacy and patient safety. We developed a rapid quality-detection workflow for CCFG integrating multi-dimensional biomimetic sensory data. Fifty-one batches of CCFG were analyzed by HPLC and GC-IMS and characterized with near-infrared spectroscopy (NIR), electronic eye (EE), electronic tongue (ET), and electronic nose (EN). We built qualitative classification models using convolutional neural networks (CNN), partial least squares discriminant analysis (PLS-DA), k-nearest neighbors (KNN), and least-squares support vector machines (LS-SVM) and quantitative component-prediction models using partial least squares regression. In qualitative models, the CNN model based on GC-IMS data achieved 100% accuracy. The multi-source information fusion of ET+EN+EE achieved the highest positive judgement rate of 97.06%. For quantitative prediction, the fusion model combining the EN+EE predicted cinnamic acid content with R² = 0.9834, RMSEC = 0.3195, and RMSEP = 0.3547, demonstrating excellent predictive performance. Overall, this approach provides a rapid, robust method for CCFG quality identification extending conventional quality-control practices for traditional Chinese medicine (TCM) and offering a practical methodological reference for assessing similar TCM formula granules and related food products.
科研通智能强力驱动
Strongly Powered by AbleSci AI