神经网络预后评价在脑胶质瘤患者术后生存质量分析中的应用  

Application of artificial neural networks in the evaluation on quality of life of patients with Glioma after surgery

作　　者：李少琼[1] 赵晓军[2] 申宁宁[1] 高宇钊 萨建[1] 刘桂芬[1] 

机构地区：[1]山西医科大学公共卫生学院流行病与卫生统计学教研室,山西太原030001 [2]山西省肿瘤医院神经外科

出　　处：《中国预防医学杂志》2015年第3期175-179,共5页Chinese Preventive Medicine

基　　金：国家自然科学基金项目(81172774);国家青年科学基金项目(81102198);太原市科技计划项目大学生创新创业专题(120164023)

摘　　要：目的构建COX比例风险预测模型与人工神经网络预测模型,对脑胶质瘤患者术后生存质量进行评价,为临床医师提供简单、准确的评估方法。方法收集2010年6月至2013年8月山西省肿瘤医院收治的58例脑胶质瘤患者的住院治疗及随访资料的年龄、性别、职业等人口学特征,患者入院时的症状、体征、核磁共振成像(magnetic resonance imaging,MRI)检查、病理诊断分型等,肿瘤切除程度、免疫组化检查及Karnofsky功能状态(Karnofsky performance status,KPS)评分等,筛选有意义因素,建立COX比例风险模型,采用预后指数分层和人工神经网络模型,预测患者术后1年生存质量;并采用ROC分析,对两种方法的预测能力进行评价。结果 COX比例风险模型分析表明,伴有癫痫、术前KPS评分、KI67、病理级别、肿瘤切除程度、血供、肢体活动障碍是影响脑胶质瘤患者术后生存质量的主要影响因素。COX比例风险预测模型的灵敏度为60.0%,特异度为83.3%;人工神经网络预测模型的灵敏度为80.0%,特异度为83.3%。结论人工神经网络模型的预测效果优于COX比例风险模型,人工神经网络可为临床医师评价脑胶质瘤患者术后生存质量提供个体化治疗方法。Objective To establish a COX proportional hazards model and artificial neural networks model to evaluate the quality of life of patients with glioma after surgery,and to provide a simple predictive way for clinicians. Methods Data of 58 patients with glioma hospitalized for surgery in Shanxi Tumor Hospital from June 2010 to August 2013 were retrieved and analyzed.COX proportional hazards model and artificial neural networks model were established to predict the quality of life of patients.Recipient curve method（ROC）analysis was used to compare these two models. Results The result of COX proportional hazards model showed that epilepsy,KPS score before surgery,KI67,age,pathological grade,removal of tumor,blood supply,limb movement disorder were major influencing factors.The sensitivity and specificity of COX proportional hazards model were 60% and 83.3%,while,they were 80% and 83.3%for artificial neural networks model. Conclusions Artificial neural networks model is superior to COX proportional hazards model in the prediction of quality of life of patients with glioma after surgery.Therefore,artificial neural networks model is able to help clinician to treat patients individually in order to improve the quality of life of patients.

关 键 词：脑胶质瘤 生存质量 预后指数 人工神经网络 预后评价 

分 类 号：R739.41[医药卫生—肿瘤]