论文部分内容阅读
目的:评价年龄、前列腺特异性抗原(PSA)以及经直肠前列腺超声影像特征构建的TAN贝叶斯网络(tree-augmented Nave Bayesian network)模型对前列腺癌的预测效果。方法:收集2008年1月至2011年9月行前列腺穿刺活检941例患者的临床数据,包括年龄、PSA、超声影像以及病理诊断,构建TAN贝叶斯网络,对前列腺癌进行预测,并与病理诊断“金标准”比较。结果:941例患者中,358例经活检证实为前列腺癌,583例为非前列腺癌性病变。TAN贝叶斯网络对前列腺癌预测的准确率为85.11%、灵敏度88.37%、特异性83.67%、阳性预测值70.37%、阴性预测值94.25%。结论:基于年龄、PSA以及经直肠前列腺超声影像构建的TAN贝叶斯网络模型对前列腺癌预测效果较好,可作为临床筛查或诊断前列腺癌的一种方法。
OBJECTIVE: To evaluate the predictive value of age, prostate-specific antigen (PSA), and TAN-Bayesian network model built by transrectal ultrasound imaging features in prostate cancer. Methods: The clinical data of 941 patients undergoing prostate biopsy from January 2008 to September 2011 were collected, including age, PSA, ultrasonography and pathological diagnosis. The TAN Bayesian network was constructed and the prognosis of prostate cancer was predicted. Diagnosis “gold standard ” comparison. Results: Of the 941 patients, 358 were biopsy-proven prostate cancer and 583 were non-prostate cancer lesions. The accuracy of TAN Bayesian network in predicting prostate cancer was 85.11%, sensitivity 88.37%, specificity 83.67%, positive predictive value 70.37% and negative predictive value 94.25%. CONCLUSION: The TAN Bayesian network model based on age, PSA and transrectal ultrasound imaging of prostate is good for predicting prostate cancer and can be used as a clinical screening or diagnosis of prostate cancer.