摘要
目的利用深度学习算法建立前列腺癌生化复发预测模型,为前列腺癌根治术后患者早发现、早诊断、延长患者生存期提供参考。方法收集2001年3月—2016年11月北京大学第一医院泌尿外科接受前列腺癌根治术的442例患者的临床信息作为变量,应用五折交叉验证法将其划分为训练集(n=412)和验证集(n=30),采用深度学习算法(CNN-BiLSTM、CNN-LSTM、BiLSTM、CNN-BiGRU)建立前列腺癌生化复发预测模型,其中验证集用于评估模型性能和临床应用的可能性。结果在4种深度学习的算法中,CNN-BiLSTM算法准确率最高为76.7%,受试者工作曲线下面积为0.71。结论基于前列腺癌根治术后患者的多种临床信息,通过深度学习方法建立前列腺癌生化复发预测模型具有较高的准确率,能够为预测前列腺癌的生化复发提供一定参考。
Objective To establish a prediction model for biochemical recurrence of prostate cancer based on deep learning algorithm,and to provide reference for the early detection,early diagnosis and prolonged survival of patients after radical prosta⁃tectomy.Methods Clinical data of 442 patients who received radical prostatectomy in the Department of Urology,Peking Uni⁃versity First Hospital from Mar.2001 to Nov.2016 were collected.The patients were divided into training set(n=412)and vali⁃dation set(n=30)with five-fold cross-validation method.Deep learning algorithms inducing CNN-BiLSTM,CNN-LSTM,BiLSTM and CNN-BiGRU were used to establish a prediction model for the biochemical recurrence of prostate cancer.The vali⁃dation set was used to evaluate the performance of the model and possibility of clinical application.Results Among the 4 deep learning algorithms,CNN-BiLSTM algorithm had the highest accuracy of 76.7%,and the area under the receiver operating char⁃acteristic curve was 0.71.Conclusion Based on the clinical data of patients after radical prostatectomy,the prediction model of biochemical recurrence of prostate cancer established by deep learning method has high accuracy,which can provide certain ref⁃erence for predicting biochemical recurrence of prostate cancer.
作者
高文治
何宇辉
夏漫城
巩艳青
何世明
张建烨
周利群
郭跃先
李学松
GAO Wenzhi;HE Yuhui;XIA Mancheng;GONG Yanqing;HE Shiming;ZHANG Jianye;ZHOU Liqun;GUO Yuexian;LI Xuesong(Department of Urology,The First Medical Hospital of Peking University,Beijing 100000;Department of Urology,The Third Hospital of Hebei Medical University,Shijiazhuang 050000,China)
出处
《现代泌尿外科杂志》
CAS
2022年第3期230-233,共4页
Journal of Modern Urology
关键词
前列腺癌
生化复发
深度学习
预测模型
prostate cancer
biochemical recurrence
deep learning
prediction model
