摘要
目的应用基于链式方程的填补方法处理医学研究中的数据缺失,并以填补后完全数据构建联合指标的logistic判别函数,判断其在前列腺癌的预测诊断中的应用价值。方法采用模拟研究,针对现实数据缺失情况模拟不同填补集结果,并以此对现实数据进行填补,以完整数据构建logistic判别,进行分析预测。结果填补结果随着填补次数的增加而逐渐接近真实值并趋于稳定。联合年龄、血清前列腺特异性抗原值、血流阻力指数及经直肠前列腺超声检查指标的logistic判别分析结果的灵敏度为82.39%,特异度为74.86%。结论联合指标分析可提高前列腺癌的诊断预测水平,以减轻患者穿刺痛苦。
Objective To predict the application value of logistic discriminant analysis in prostate cancer diagnosis by constructing logistic discriminant function using multiple imputation by chained equation to deal with missing data in medical research.Methods In view of the the present missing data,we simulated different imputed results to impute the data,and then constructed logistic discriminant analysis for predicting.Results More imputations,closer the results were to the true value,and the results also tended to be stable.Combined with age,serum PSA,resistance index and transrectal prostatic ultrasonography index,the sensitivity of logistic discriminant analysis results was 82.39%,and specificity was 74.86%.Conclusion Combined index analysis can improve the diagnosis of prostate cancer prediction level,so as to alleviate the puncture pain of patients.
出处
《华西医学》
CAS
2013年第2期200-203,共4页
West China Medical Journal
基金
四川省科技厅应用基础研究(2009JY0020)~~
