摘要
目的应用偏最小二乘判别分析(partial least-squares discriminant analysis,PLSDA)法评估九一丹外用治疗浆细胞性乳腺炎的安全性及其临床价值。方法观察50例九一丹外用治疗浆细胞性乳腺炎患者在用药前、用药后第4天、停药第1、14天检测的血尿生化指标;在用药前、用药后第1、4、7天、停药第1、14天检测患者血汞、尿汞,并于停药第28天和停药3个月检测患者尿汞含量。并在用药前、用药后第1、4、7天及停药第1天记录创面情况,用药量及用药天数。在筛选到有重要影响的安全性指标后,PLSDA建立影响因素的潜在安全性判别模型,采用ROC曲线下面积评估模型的适用性,并使用变量重要性投影(variable importance in the projection,VIP)评价潜在的重要影响因素。结果九一丹外用治疗浆细胞性乳腺炎对尿β2-微球蛋白(β2-MG)、尿N-乙酰-β氨基葡萄糖苷酶(NAG)、24 h尿蛋白及α1微球蛋白(α1-MG)影响较大;PLSDA判别模型预测准确率为74.00%,模型的特异度、灵敏度及ROC曲线下面积分别为0.7826、0.7037及0.8084。通过模型筛选到3个对潜在安全性具有较大影响的因素:用药前疮腔体积、用药天数和用药量。结论PLSDA法能够用于分析中医临床的相关信息,筛选出九一丹外用安全性监测的主要指标为尿β2-MG和尿NAG;用药量和用药时间是影响血汞、尿汞的最主要因素,依据这两大因素建立的九一丹治疗浆细胞性乳腺炎安全性分级仿真模型能在一定程度上评估临床九一丹外用的安全性。
Objective To evaluate the safety and the clinical value of external use of Jiuyi Pow- der （JP） in treating plasma cell mastitis using partial least-squares discriminant analysis （PLSDA）. Methods Totally 50 patients with plasma cell mastitis treated by external use of JP were observed and biochemical examinations of blood and urine detected before application, at day 4 after application, at day 1 and 14 after discontinuation. Blood mercury and urinary mercury were detected before application, at day 1,4, and 7 after application, at day 1 and 14 after discontinuation. Urinary mercury was also de- tected at 28 after discontinuation and 3 months after discontinuation. The information of wound, days of external application and the total dosage of external application were recorded before application, at day 1, 4, and 7 after application, as well as at day 1 after discontinuation. Then a discriminant model covering potential safety factors was set up by PLSDA after screening safety indices with important effects. The applicability of the model was assessed using area under ROC curve. Potential safety factors were as-sessed using variable importance in the projection （VIP）. Results Urinary β2-microglobulin （β2-MG）, u- rinary N-acetyl-β-D-glucosaminidase （NAG）, 24 h urinary protein, and urinary α1-microglobulin （α1-MG） were greatly affected by external use of JP in treating plasma cell mastitis. The accuracy rate of PLSDA discriminate model was 74.00%. The sensitivity, specificity, and the area under ROC curve was 0.7826, 0.7037, and 0.8084, respectively. Three factors with greater effect on the potential safety were screened as follows ：pre-application volume of the sore cavity, days of external application, and the total dosage of external application. Conclusions PLSDA method could be used in analyzing bioinformation of clinical Chinese medicine. Urinary β2-MG and urinary NAG were two main safety monitoring indices. Days of ex- ternal application and the total dosage of external applica
出处
《中国中西医结合杂志》
CAS
CSCD
北大核心
2015年第4期429-433,共5页
Chinese Journal of Integrated Traditional and Western Medicine
基金
顾氏外科流派传承研究(No.ZYSNXD-CC-APGC-JD002)
国家"十一五"科技支撑计划基金资助项目(No.2008BAI53B083)
关键词
浆细胞性乳腺炎
九一丹
外治法
模型
偏最小二乘判别分析
plasma cell mastitis
Jiuyi Powder
external therapy
model
partial least-squares discriminant analysis
