基于機器學(xué)習(xí)算法構(gòu)建老年重癥病人靜脈血栓栓塞癥風(fēng)險預(yù)測模型

打開文本圖片集

Development of risk prediction model for venous thromboembolism in elderlycriticallyill patient based on machine learning

JIN Jiel， XU Qing2，LU Jie'， ZHAO Jiayue'，ZHANG Qing1，KONG Yang1* ， XU Hongmeil* 1.Binzhou Medical Colege，Shandong 25660o China;2.Binzhou Medical College Yantai Affiliated Hospital *Corresponding AuthorKONG Yang，E-mail： kongyang@ bzmc.edu.cn; XU Hongmei，E-mail： hmx58@163.com

AbstractObjective：Todeveloprisk prediction modelforvenous thromboembolism（VTE）inelderlycriticalyillpatients basedon machineleatotaflillltoeUtee hospitalsinShandongprovincewereselectedasstudysubjectsfromJanuary2O2OtoJune2O23.Andclinicaldatawerecolected.The patients wererandomlydivided intotrainigset36cases）andvalidationset273cases）at7：3ratio.heocurrenceofVTEduringICU hospitalizationwasused as theoutcome variable.Predictionmodels wereconstructed using 4machinelearing，namelyrandomforest， extremegradientbosting，supportvectormachines，andgadientbostigdecision tre.Modelperformancewas evaluatedusingmetrs suchasareaunderthecure（AUC）ofreceiveoperatorcharacteristicandBrierscoreandteoptialmodelwasselected.Inteeability analysisfthebestperformingmodelasconductedusingteSHAPlgorithmResults：Amongthe909elderlyrticallillpatients58 developed VTE， with incidence of 28.4% .Among the 4 models，the random forest achieved the higher AUC（O.8O3），accuracy（0.733）， senitivity（.662），ndspecificiy（0.76），alongiththelowestBrierScore（O.171）.onclusions：TheriskpredictionmodelforE inelderlycriallyillpatientsdevelopedbasedonandomforestdemonstratedstrongpredictiveperformance.Itouldproideeference for optimizing VTE management in elderly critically ill patients.

Keywordsmachine learming; elderly;critical illess;venous thromboembolism; prediction model; random forest

doi：10.12102/j.issn.1009-6493.2025.14.002

靜脈血栓栓塞癥（venousthromboembolism，VTE）包括深靜脈血栓形成（deepvenousthrombosis，DVT）和肺栓塞（pulmonarythromboembolism，PE），是僅次于心肌梗死和腦卒中的第三大常見心血管疾病，具有發(fā)病率高、病死率高以及復(fù)發(fā)率高的特點，是重癥監(jiān)護室（intensivecareunit，ICU）病人常見的并發(fā)癥之一[1-2]。（剩余11466字）