Machine learning methods to predict the cultivation age of Panacis Quinquefolii Radix

• DOI: 10.1186/s13020-021-00511-5
• Publication Date: 2021-10-10T20:05:27Z
• Authors: Xiaowen Hu; Hua Yan; Xiaodong Wang; Zonghu Wang; Yuanpeng Li; Lianjun Zheng; Jianbo Yang; Wenguang Jing; Xianlong Cheng; Feng Wei; Shuangcheng Ma
• Keywords: Radix (gastropod)

American ginseng (AG) is a valuable medicine widely consumed as herbal remedy throughout the world. Huge price difference among AG with different growth years leads to intentional adulteration for higher profits. Thus, developing reliable approaches authenticate cultivation ages of products great use in preventing age falsification.A total 106 batches samples along their 9 physicochemical features were collected and measured from experiments, which was then split into training set two test sets (test 1 2) according regions. Principle component analysis (PCA) carried out examine distribution three data sets. Four machine learning (ML) algorithms, namely elastic net, k-nearest neighbors, support vector multi-layer perception (MLP) employed construct predictive models using inputs outputs. In addition, similarity-based applicability domain (AD) defined these ensure reliability results produced regions.A positive correlation observed between several years. PCA revealed diverse distributions The most accurate model derived MLP shows good prediction power fivefold cross validation mean square error (MSE) 0.017 0.016 respectively, but MSE value 1.260 2. After applying AD, all showed much lower errors within AD (IDs) than those outside (ODs). remains best an 0.030 IDs.Cultivation have close relationship bioactive components AG. constructed are also able predict discriminate that inaccurate results. AD-equipped used this study provide useful tools determining market freely available at https://github.com/dreadlesss/Panax_age_predictor .