老年髋部骨折患者术前深静脉血栓形成的机器学习预测模型构建

打开文本图片集

中图分类号 R683 文献标识码 A文章编号 2096-7721（2025）06-1019-06

Development a machine learning prediction model for preoperative deepvein thrombosisin elderlyhipfracture patients

HUANG Yufeng QIU Yanran

AbstractObjective：Toidentifyriskvariablesfor preoperativedeepvein thrombosis （DVT）formation inelderly patients withhip fractures usingmachinelearingalgorithmsandtoconstructanomogrammodelforoptimizingpreoperativeasessenandpreventive measures.Methods：174elderlypatients withhipfracturesfromJune2023 toDecember2024 wereenroledand divided ito the thrombus group（n=57）andthenon-thrombus group （n=117）basedonpreoperativelower extremityvenous ultrasonography.Two machineearningalgorithms，LeastAsoluteShikageandSelectionOperatorLASO），ndSupportVectorMacine-RecusieFature Elimination（SE）sellauivaeListicgordtreeaisbl.li highestAUCwasselectedtoosrctheomoam，flowdbyalidatio.Results：MultivarateLgistcegesioidentifdtro injutt systemicmmune-inflammationindex（SII）asindependentiskfactorsfrpreoperativeDVT.TheLASSOmodel generatedineatures withnon-zerontsuigtfrjutspialdiss，W，，Nroe-te ration（MLR），LR，ndSeValgidentifdfisribs，cuingtifrojutospitadi D-dimer，FIB，ndSIROCanalissoedtatthAUCforLSOSVFE，andLgisigressnere069，97d93， respectively. The SVM-RFE model had the highest sensitivity （98.18 % ） and specificity （90.77 % ）. The SVM-RFE model was selected to construtheomogam，andtheHosmer-Lemeshowgoodness-fitestdemonstratedgoodcalirationabilitythenogam χ2 =2.157， （20 P =0.867）.ConcusionTheconstructedomogamodelcanasisthaltcareproviders inaccuratelyassessgthepreopertiveDV riskinelderlypatitsihifrcturs，tebyabgtielypreventivemeasues，tiingtealoatiodicalsod improving the quality care.

KeyWordsHipFracture;Deep Vein Thrombosis;Machine Learning;Nomogram Model; Risk Prediction.

髋部骨折是老年人常见的骨骼损伤，其发病率随着人口老龄化的加剧而不断上升。（剩余9504字）