论文部分内容阅读
目的应用支持向量机算法(SVM)分析职业性三氯乙烯药疹样皮炎(OMLDT)患者血清中差异表达miRNAs的相对表达水平,并建立OMLDT诊断模型。方法收集2009年1月1日至2014年12月30日经诊断的OMLDT患者和接触者血清样本,应用miRNA芯片筛选差异表达的miRNA并进行qRT-PCR验证。从全部样本中随机抽取60%作为测试集应用SVM建立OMLDT诊断模型,余下40%作为验证集评价模型的特异性和灵敏度。结果本研究中OMLDT病例组34例,其中男性23例、女性11例,平均年龄为(27.29±10.43)岁。对照组34人,其中男性26人、女性8人,平均年龄为(25.00±6.14)岁。2组人群在性别构成和平均年龄的差异上均无统计学意义(均P>0.05)。miR-21和miR-193b在OMLDT组中的表达量高于对照组(均P<0.01)。构建诊断模型的交叉验证识别率为95%,模型灵敏度和特异性分别为81.82%、100.00%,总体预测正确率为92.86%。结论 miR-21和miR-193b可能是OMLDT患者外周血中潜在的生物标志。基于SVM算法建立的OMLDT诊断模型拟合效果好,可为临床早期诊断提供有价值的线索。
Objective To analyze the relative expression levels of differentially expressed miRNAs in serum of patients with occupational trichlorethylene eruptive dermatitis (OMLDT) using Support Vector Machine (SVM) and establish a diagnostic model of OMLDT. Methods Serum samples of OMLDT patients and contacts diagnosed from January 1, 2009 to December 30, 2014 were collected. The differentially expressed miRNAs were screened by miRNA microarray and verified by qRT-PCR. 60% of the samples were randomly selected from all the samples as the test set to establish the OMLDT diagnostic model using SVM. The remaining 40% was used as a validation set to evaluate the specificity and sensitivity of the model. Results In this study, 34 cases of OMLDT patients, including 23 males and 11 females, with an average age of (27.29 ± 10.43) years. The control group of 34 people, including 26 males and 8 females, with an average age of (25.00 ± 6.14) years. There was no significant difference in gender composition and mean age between the two groups (all P> 0.05). The expression of miR-21 and miR-193b in OMLDT group was higher than that in control group (all P <0.01). The recognition rate of constructing cross-validation of the diagnosis model was 95%, the sensitivity and specificity of the model were 81.82% and 100.00% respectively, and the overall prediction accuracy rate was 92.86%. Conclusion miR-21 and miR-193b may be potential biomarkers in peripheral blood of OMLDT patients. The OMLDT diagnosis model based on SVM algorithm has good fitting effect and can provide valuable clues for early clinical diagnosis.