数学模型在良、恶性腹水鉴别诊断中的应用

【目的】建立并运用数学模型来定量探讨多种指标联合检测对良、恶性腹水的鉴别价值。【方法】对历史数据进行单因素和多因素Logistic逐步回归得出预测腹水性质的数学模型即回归方程，借助ROC曲线评价其预测效率，并和腹水细胞学结果比较以评价其诊断效能。【结果】①单因素分析显示有统计学意义的指标有sAAG，血清CA199、AFP、CEA，腹水CEA。②多因素分析进入回归方程的指标有sAAG，血清CAl99、AFP和腹水CEA，得出的回归方程为P=1／1＋e-（-1．417—0．745x1＋0．060x2＋0．786x3＋1．932x7）。③回归方程对腹水良、恶性预测有显著统计学意义。【结论】所建立的预测模型和诊断腹水的金标准腹水病理细胞学检测相比具有较好的预测价值，大大提高了诊断的科学性和准确性。

Application of Mathematical Model in Differential Diagnosis of Benign and Malignant Ascites

[Objective]To establish a mathematic model to explore the value of combination detection of many markers for the differential diagnosis of benign and malignant ascites. [Methods]After the historical data were analyzed with univariate and multivariate logistic regression, the mathematical model（regression equa- tion） for the differential diagnosis of benign and malignant ascites had been deduced. The predictive efficacy of the model was evaluated by ROC curve and was compared with the results of cytology. [Results] Univariate a nalysis showed that CA199, AFP and CEA had statistical significance. Multivariate analysis showed that SAAG, CA199, AFP and CEA were the indicators in regression equation. The regression equation, P= 1/1＋ e-（ 1. 417--0. 745x1＋0. 060x2＋0. 786x3q 1. 932x7） was deduced. The sensitivity, specificity and veracity of the combination of the useful items were 92.3%,100% and 96.6%, respectively in diagnosis of malignant ascites. The regression equation for the differential diagnosis of benign and malignant ascites had statistical significant. [Conclusion]The mathematical model for the differential diagnosis of benign and malignant ascites has better value than the golden standard by pathological cytological method. It can increase the science and accurate of diagnosis.