良恶性腹水的鉴别及数学模型的建立

发布时间：2018-05-26 17:54

本文选题：腹水 + 肿瘤标记物　；参考：《兰州大学》2017年硕士论文

【摘要】：背景及目的:腹水是临床常见的一种病征,腹水可分为良性腹水及恶性腹水,良性腹水又可分为结核性腹水及非结核性良性腹水。腹水良恶性的鉴别与临床治疗方案的选择及其预后息息相关,但是部分腹水患者良恶性的鉴别在临床中仍然很困难。目前最常用的方法仍是脱落细胞学检查,虽具有较高的诊断特异性,但是灵敏度、准确性较低,在临床上的应用受到限制。现急需一种简单易行、准确率较高的诊断方法对腹水性质进行鉴别。临床中通过单项实验室指标的检测对腹水的性质进行诊断其准确率仍低,多项指标共同检测可提高其诊断的准确率,但过程较为复杂。本文通过对临床病例进行回顾性研究,经单因素分析筛选出对腹水性质鉴别诊断有意义的实验室指标,再通过多因素Logistic回归分析创建数学诊断模型。将入选的实验室指标作为一整体,用该数据定量来反映疾病发生可能性,以提高诊断的灵敏度、特异性和准确性。方法:本文采用回顾性分析,对2010年1月至2016年12月在兰州大学第一医院消化内科住院的172例腹水患者的临床病例资料进行分析,收集患者血清AFP、CEA、CA125、CA199、CA724、LDH、GLU、ALP、GGT、TP、ALB,腹水 AFP、CEA、CA125、CA199、CA724、LDH、GLU、ALP、GGT、TP、ADA, TP之差共23项实验室指标,先对这些指标进行单因素分析,将有统计学意义的实验指标作为白变量,分别以良、恶性腹水及结核、非结核性腹水为因变量,再进行多因素logistic回归分析,分别建立针对恶性腹水和结核性腹水的诊断预测数学模型,并用ROC曲线对其诊断效能进行分析。结果:1.引起腹水常见的三大病因依次为肝硬化失代偿期、恶性肿瘤和结核性腹膜炎。2.恶性腹水患者发病年龄要高于良性腹水患者,女性患病率多于男性;结核性腹水患者发病年龄低于非结核性腹水患者,男女的患病率无明显差异。3.对良恶性腹水的鉴别:针对恶性腹水我们建立的诊断方程为:P1 = 1/[1 + e~(-(- 3.859 + 0.082X1+0.001X2+ 0.003X3))]其中X1 腹水 CEA,X2=腹水CA125, X3=腹水LDH,P1为预测概率,e为自然对数。数学诊断模型的ROC曲线下面积为0.944,最佳临界值为0.515,灵敏度为77.78%,特异性为98.31%,准确性为91.86%,漏诊率为22.22%,误诊率为1.69%,约登指数为76.09%,阳性预测值为95.45%,阴性预测值为90.63%,阳性似然比为46.02,阴性似然比为 0.23。4.对结核性及非结核性腹水的鉴别:针对结核性腹水我们建立的诊断方程为:P2 = 1/[1 + e~(-(-7.466-0.005X1+0.104X2 + 0.010X3 + 0.130X4))] 其中X1=腹水LDH,X2=腹水TP,X3=腹水GGT,X4=腹水ADA,P2为预测概率,e为自然对数。数学诊断模型的ROC曲线下面积为0.978,最佳临界值为0.302,灵敏度为93.18%,特异性为94.53%,准确性为94.19%,漏诊率为6.82%,误诊率为5.47%,约登指数为87.81%,阳性预测值为85.42%,阴性预测值为97.58%,阳性似然比为17.03,阴性似然比为0.07。5.对恶性腹水及结核性腹水进行综合鉴别:如果Pre-10.515、Pre-20.302,高度怀疑为恶性腹水;Pre-10.515、Pre-20.302,高度怀疑为结核性腹水;Pre-10.515、Pre-20.302,则诊断为良性非结核性腹水;Pre-10.515、Pre-20.302,则比较哪个更接近于1即对其中之一的疾病诊断更符合。结论:运用Logistic回归分析建立的数学诊断模型,可将入选的实验室指标作为一整体,用定量数据对腹水性质进行鉴别,可提高诊断的灵敏度、特异性和准确性。
[Abstract]:Background and objective: ascites is a common clinical symptom. Ascites can be divided into benign ascites and malignant ascites. Benign ascites can be divided into tuberculous ascites and non tuberculous benign ascites. The identification of benign and malignant ascites is closely related to the choice of clinical treatment scheme and its prognosis, but the differentiation of benign and malignant partial ascites is still in clinical practice. It is still difficult. The most commonly used method is still exfoliative cytology, although it has high diagnostic specificity, but sensitivity, accuracy is low, and its clinical application is limited. It is urgent to identify the properties of ascites by a simple and accurate diagnostic method. The accuracy rate of the diagnosis of ascites is still low. The common detection of multiple indexes can improve the accuracy of diagnosis, but the process is more complex. This article through a retrospective study of clinical cases, through single factor analysis to screen out the significance of the laboratory indicators for the differential diagnosis of ascites properties, and then by multi factor Logistic regression analysis to create A mathematical diagnosis model. Using the selected laboratory indicators as a whole, this data is used to reflect the possibility of the disease, in order to improve the sensitivity, specificity and accuracy of the diagnosis. Methods: a retrospective analysis was adopted in this paper for 172 cases of ascites hospitalized in the digestive department of First Hospital Affiliated to Lanzhou University from January 2010 to December 2016. 23 laboratory indexes of the patients' serum AFP, CEA, CA125, CA199, CA724, LDH, GLU, ALP, GGT, TP, ALB, and ascites are analyzed in a single factor analysis, and a statistically significant experimental index is used as a white variable, with good and malignant abdomen respectively. Water and tuberculosis, non tuberculous ascites as the dependent variables, and then multifactor logistic regression analysis, the mathematical models for the diagnosis and prediction of malignant ascites and tuberculous ascites were established respectively, and the diagnostic efficiency was analyzed with the ROC curve. Results: 1. the three common diseases caused by ascites were due to the decompensation period of cirrhosis, malignant tumor and tuberculosis. The age of the patients with.2. malignant ascites is higher than that of the benign ascites. The prevalence of female is more than that of the male; the age of the patients with tuberculous ascites is lower than that of the non tuberculous ascites. The prevalence rate of male and female is not significantly different from that of the benign and malignant ascites. The diagnostic equation for malignant ascites is: P1 = 1/[1 + e~ (- (- 3.). 859 + 0.082X1+0.001X2+ 0.003X3)) X1 ascites CEA, X2= ascites CA125, X3= ascites LDH, P1 as the prediction probability, e as the natural logarithm. The area under the ROC curve of the mathematical diagnosis model is 0.944, the optimum critical value is 0.515, the sensitivity is 77.78%, the specificity is 98.31%, the accuracy is 91.86%, the missed diagnosis rate is 22.22%, the misdiagnosis rate is 1.69% and the Jordan index is 7. 6.09%, the positive predictive value was 95.45%, the negative predictive value was 90.63%, the positive likelihood ratio was 46.02, the negative likelihood ratio was 0.23.4. for the identification of tuberculous and non tuberculous ascites. We established the diagnostic equation for tuberculous ascites: P2 = 1/[1 + e~ (- (- (- (- (- (- (- (-7.466-0.005X1+0.104X2 + 0.010X3 + 0.130X4))] in which X1= ascites LDH, X2= ascites TP, X3= Ascites GGT, X4= ascites ADA, P2 as the prediction probability, e is the natural logarithm. The area of the ROC curve under the mathematical diagnosis model is 0.978, the optimum critical value is 0.302, the sensitivity is 93.18%, the specificity is 94.53%, the accuracy is 94.19%, the missed diagnosis rate is 6.82%, the misdiagnosis rate is 5.47%, the Jordan index is 87.81%, the positive predictive value is 85.42%, the negative predictive value is 97.58%, Yang negative predictive value 97.58%, Yang is 97.58%, Yang Sexual likelihood ratio is 17.03, negative likelihood ratio is 0.07.5. for malignant ascites and tuberculous ascites. If Pre-10.515, Pre-20.302, highly suspected as malignant ascites; Pre-10.515, Pre-20.302, highly suspected as tuberculous ascites; Pre-10.515, Pre-20.302, and diagnosis of benign non tuberculous ascites; Pre-10.515, Pre-20.302, is the comparison. Which is closer to 1 is more consistent with one of the disease diagnoses. Conclusion: the mathematical diagnosis model established by Logistic regression analysis can use the selected laboratory indicators as a whole and identify the properties of ascites with quantitative data, which can improve the sensitivity, specificity and accuracy of the diagnosis.
【学位授予单位】：兰州大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R442.5

【参考文献】