目的: 构建以氨基酸含量为指标的用于识别冬虫夏草不同主产区的评分卡。方法: 以需要识别的主产区样品作为正样品,其他产区的样品作为负样品,对训练集样品的氨基酸含量数据进行分箱,计算分箱的证据权重（WOE）和信息价值（IV）,调整优化至适宜的分箱数量。使用WOE对原始数据进行编码,并建立逻辑回归模型。根据评分公式设计合适的评分,以50分作为正负样品几率为1时的分数,计算基准分数和各分箱对应分数。使用基准分数与待评分样品对应各分箱的对应分数相加,便可得出该样品的分数,以50分为阈值可以对正负样品概率大小进行判别。结果: 所构建的识别青海产区冬虫夏草评分卡和识别西藏产区冬虫夏草评分卡对于测试集样品的识别准确率分别为0.85和0.90。结论: 所构建的2套评分卡能够简便、准确地识别冬虫夏草主产区。人工智能技术与药物分析相融合,可以显著地对传统药物质量分析、评价和控制起到提质增效的作用。
Objective: To construct scoring cards using amino acid contents as indicators to identify different main geographical origins of Cordyceps. Methods: Samples from the main geographical origins to be identified were used as positive samples, while samples from other areas served as negative samples. The amino acid content data of the training set samples were divided into bins, and the weight of evidence (WOE) and information value (IV) for each bin were calculated to optimize the binning strategy. The original data were encoded using WOE, and a logistic regression model was established. A suitable scoring system was designed according to the scoring formula, with 50 points set as the baseline when the probability of positive and negative samples was equal, determining the base score and the scores corresponding to each bin. By summing the base score with the scores corresponding to each bin for the samples to be scored, the final score of the sample could be derived. A threshold of 50 points could be used to discriminate the likelihood of positive and negative samples. Results: The scoring cards constructed for identifying Cordyceps sinensis from the Qinghai and Xizang production areas achieved identification accuracies of 0.85 and 0.90, respectively, for the test set samples. Conclusion: The two constructed scoring cards can simply and accurately identify the main production areas of Cordyceps sinensis. The integration of artificial intelligence technology with pharmaceutical analysis can significantly enhance the quality and efficiency of traditional drug quality analysis, evaluation, and control.
[1] 中华人民共和国药典2020年版.一部[S].2020:119
ChP 2020.Vol Ⅰ[S].2020:119
[2] 王维恩. 不同产地冬虫夏草中微量元素分析[J].光谱学与光谱分析,2023,43(10):3247
WANG WE.Analysis of trace elements in Ophiocordyceps sinensis from different habitats[J].Spectrosc Spect Anal,2023,43(10):3247
[3] 卫秋阳,邓小书,贺元川,等.冬虫夏草产区土壤特征和微生物组成分析[J].环境昆虫学报,2023,45(6):1559
WEI QY,DENG XS,HE YC,et al.Analysis of the specificity soil characteristics and microbial composition in Ophiocordyceps sinensis[J].J Environ Entomol,2023,45(6):1559
[4] 陈建博,李秀璋,徐成体,等.青海省冬虫夏草采挖区与非采挖区土壤生态化学计量特征[J].草地学报,2023,31(4):1134
CHEN JB,LI XZ,XU CT,et al.Soil ecological stoichiometry in the excavated and non-excavated areas of Chinese Cordyceps in Qinghai province[J].Acta Agrestia Sin,2023,31(4):1134
[5] 张灵迂,周罗静,钟洪金,等.微生物多样性对冬虫夏草形成及品质的影响[J].中药材,2023,46(6):1563
ZHANG LY,ZHOU LJ,ZHONG,HJ,et al.The impact of microbial diversity on the formation and quality of Cordyceps sinensis[J].J Chin Med Mater,2023,46(6):1563
[6] 邓大松,赵玉龙.我国商业银行小微企业申请评分卡构建及验证研究[J].投资研究,2017,36(5):149
DENG DS,ZHAO YL.Research on the application card on small enterprise in the commercial bank[J].Rev Invest Stud,2017,36(5):149
[7] 邵永运,张立莹.制造业上市公司财务数据异常风险评分卡模型[J].沈阳师范大学学报(自然科学版),2023,41(6):556
SHAO YY,ZHANG LY.Financial data anomaly risk scorecard model of listed manufacturing companies[J].J Shenyang Norm Univ(Nat Sci Ed),2023,41(6):556
[8] 尚长春,雷璐华,许远杏,等.一种基于评分卡的健康测度方法:以桂林市民健康调查为例[J].健康研究,2021,41(3):251
SHANG CC,LEI LH,XU YX,et al.A health measurement method based on score card:a case study of Guilin residents’ health survey[J].Health Res,2021,41(3):251
[9] 梁雪枫,安婧,金娜,等.甘肃省2019—2020年出生儿童预防接种行为评分卡模型构建[J].中国疫苗和免疫,2022,28(4):465
LIANG XF,AN J,JIN N,et al.Construction of a behavior scorecard model for vaccination of children born in 2019—2020 in Gansu province[J].Chin J Vaccine Immun,2022,28(4):465
[10] 石岩,魏锋,王钢力,等.自组织映射神经网络用于氨基酸为特征的冬虫夏草道地产区识别研究[J].中国中药杂志,2021,46(18):4765
SHI Y,WEI F,WANG GL,et al.Identification of geographical origins of Cordyceps based on data of amino acids with self-organizing map neural network[J].China J Chin Mater Med,2021,46(18):4765
[11] 石岩,李宁,魏锋.机器学习算法在不同形态浙贝母与湖北贝母的干法REIMS指纹图谱鉴别分析中的应用研究[J].药物分析杂志,2024,44(1):134
SHI Y,LI N,WEI F.Research of machine learning in the application of authenticity discrimination of Fritillariae Thunbergii Bulbus and Fritillariae Hupehensis Bulbus in different form with dry-process REIMS fingerpring[J].Chin J Pharm Anal,2024,44(1):134
[12] 陈丽云,祁真.冬虫夏草氨基酸成分的药理作用分析[J].中国卫生工程学,2018,17(5):675
CHEN LY,QI Z.Analysis of amino acid composition and the pharmacology of Cordyceps sinensis[J].Chin J Public Health Eng,2018,17(5):675
[13] 林京龙,莫凡洋.人工智能赋能色谱技术研究[J].科学通报,2025,70(4):481
LIN JL,MO FY.AI-enabled chromatography research[J].Chin Sci Bull,2025,70(4):481
[14] 唐乾元. 物理学与人工智能的连接:2024年诺贝尔物理学奖解析[J].科学通报,2025,70(10):1413
TANG QY.The connections between physics and AI:a review of the 2024 Nobel Prize in physics[J].Chin Sci Bull,2025,70(10):1413
[15] 李淹博,江俊,罗毅.面向分子科学的数据智能[J].科学通报,2023,68(17):2184
LI YB,JIANG J,LUO Y.Data intelligence for molecular science[J].Chin Sci Bull,2023,68(17):2184
