髋关节骨折术后小腿肌间静脉丛血栓FP-Growth预测模型创建和验证  

Establishment and validation of FP-Growth prediction model for muscular calf vein thrombosis after hip fracture surgery

作　　者：蒋雨宸 曹海涛[2] 李劼 韩晓庆 汪彦辉 张盼盼 赵文国 JIANG Yuchen;CAO Haitao;LI Jie;HAN Xiaoqing;WANG Yanhui;ZHANG Panpan;ZHAO Wenguo(Department of Orthopedics,North China University of Science of Technology Affiliated Hospital,Hebei Province,Tangshan063000,China;Department of Orthopedics,Tangshan Second Hospital,Hebei Province,Tangshan063000,China;College of Marxism,North China University of Science and Technology,Hebei Province,Tangshan063000,China)

机构地区：[1]华北理工大学附属医院骨科,河北唐山063000 [2]河北省唐山市第二医院骨科,河北唐山063000 [3]华北理工大学马克思主义学院,河北唐山063000

出　　处：《中国医药导报》2024年第31期37-41,共5页China Medical Herald

基　　金：河北省医学科学研究课题(20201241)。

摘　　要：目的创建髋关节骨折术后小腿肌间静脉丛血栓(MCVT)FP-Growth预测模型,评价模型的预测效能。方法回顾性分析华北理工大学附属医院2014年1月至2024年1月800例髋关节骨折术后患者的临床资料。采用抽签法按照7∶3的比例将患者分为建模组(560例)和验证组(240例)。采用FP-Growth算法扫描建模组临床资料集合,基于有效强关联规则创建髋关节骨折术后MCVT预测模型。校准曲线和临床决策曲线分析进行FP-Growth预测模型内部验证;受试者操作特征曲线进行FP-Growth预测模型外部验证。结果FP-Growth算法获得频繁项集合总数为7527项,确定有效强关联规则8项。具备单独前项临床资料的患者MCVT发生率为51%~58%;具备二、三前项临床资料的患者MCVT发生率为69%~72%;具备四项临床资料的患者MCVT发生率为76%。建模组预测MCVT的C-index为0.852,预测值同实际值一致性较为理想,模型能够提供临床净收益。建模组预测MCVT的曲线下面积(AUC)为0.873,验证组预测MCVT的AUC为0.864,建模组与验证组预测MCVT的AUC比较,差异无统计学意义(P>0.05)。结论模型对髋关节骨折术后MCVT的预测效能较为理想,其预测方法和结果能够为MCVT的防治提供一定的参考。Objective To establish a FP-Growth prediction model for muscular calf vein thrombosis(MCVT)after hip fracture surgery,and to evaluate the prediction performance of the model.Methods The clinical data of 800 patients after hip fracture surgery in the North China University of Technology Affiliated Hospital from January 2014 to January 2024 were retrospectively analyzed.The patients were divided into modeling group(560 cases)and validation group(240 cases)according to a ratio of 7:3 by drawing lots.The FP-Growth algorithm was used to scan the clinical data set of the modeling group,and the prediction model for MCVT after hip fracture surgery was created based on the effective strong association rules.Calibration curves and clinical decision curve analysis were analyzed for internal validation of the FP-Growth prediction model;and external verification of FP-Growth prediction model was carried out by receiver operating characteristic curve.Results The total number of frequent item sets obtained by FP-Growth algorithm was 7527,and eight effective strong association rules were determined.The incidence of MCVT in the patients with signal first item of clinical data ranged was 51%-58%;the incidence of MCVT in the patients with two to three first items of clinical data was 69%-72%;and the incidence of MCVT in the patients with four first item of clinical data was 76%.The C-index of MCVT predicted for the modeling group was 0.852;the predicted value was in good agreement with the actual value,and the model can provide clinical net benefit.The area under the curve(AUC)of MCVT predicted by the modeling group was 0.873,and the AUC of MCVT predicted by the verification group was 0.864.There was no significant difference in AUC of MCVT prediction between modeling group and verification group(P>0.05).Conclusion The model is effective in predicting MCVT after hip fracture surgery,and its prediction method and results can provide some reference for the prevention and treatment of MCVT.

关 键 词：髋关节骨折 小腿肌间静脉丛血栓 FP-GROWTH算法 预测 

分 类 号：R683.3[医药卫生—骨科学]