摘要
目的:寻找一种准确、有效、简单易行的预测性诊断妊高征的方法。方法:用1:1配比的病例对照研究方法,收集妊高征发生的可能影响因素,采用逐步判别分析法进行分析。结果:对13个自变量通过逐步判别分析,筛选出5个有统计学意义的变量构建判别模型,用此模型对180个观察对象作回顾性评价,其灵敏度为94.4%,特异度为91.1%,阳性预测值为91.4%,阴性预测值为94.3%,误判率为7.2%;还用此模型对40个观察对象作前瞻性评价,其灵敏度为95.0%,特异度为90.0%,阳性预测值为90.5%,阴性预测值为94.7%,误判率为7.5%。结论:逐步判别分析方法可望成为一种准确、有效、简单易行的预测性诊断妊高征的工具。
Objective： To seek a accurate, effective, simple and economic method for predicting occurrence of pregnancy induced hypertension. Method Information on influence factors were collected with 1：1 matched case-control study. Date was analyzed by stepwise discriminant analysis. Results： A Fisher＇s function on predicting occurrence of pregnancy induced hypertension was established by stepwise discriminant analysis ,which was including 5 variables out of 18. One hundred and eighty individuals were discriminated retrospectively by the function with sensitivity of 94. 4%, specificity of 91.1%, positive predictive value of 91. 4%, negative predictive value of 94. 3% and error rate of 7. 2%. In addition, forty individuals were discriminated prospectively by the function with sensitivity, specificity, positive predictive value, negative predictive value, error rate with results of 95. 0%, 90. 0%, 90. 5%, 94. 7% and 7. 5% respectively. Conclusion： Stepwise discriminant analysis might serve as accurate, effective, simple and economic tool for predicting occurrence of pregnancy induced hypertension.
出处
《数理医药学杂志》
2005年第6期620-622,共3页
Journal of Mathematical Medicine
基金
广西科学研究与技术开发项目(桂科基0161008)
关键词
逐步判别分析
预测
妊高征
stepwise discriminant analysis
prediction
pregnancy induced hypertension
