基于SMOTE算法的急性缺血性脑卒中患者发生吞咽功能障碍风险预警模型的构建

打开文本图片集

中图分类号：R743.3 文献标识码：A文章编号：1006-1959（2025）12-0006-06

DOI：10.3969/j.issn.1006-1959.2025.12.002

ConstructionofaRiskWarningModel forSwallowingDysfunctioninPatientswith AcuteIschemic StrokeBased on SMOTE Algorithm

ZHANG Jia，XU Jing

（DepartmentofurologythecondAlatedHositalofjngdicalUversityUrumio，ing，a）

Abstract：ObeeTalhisctorsfalogfuciisutestroedtablisndidlag modelbasedonSOTElgoritMetodsAtotalof187patientswithacutescheicstrokeadmitedtoourhospitalfromJauarytoDecember 2022wereseeedAcodngttreatitsadspgiatddditosggoun=6ndallogfuc group（n=17）Uatedistcsalsefoolialataofoosofit independent risk factors fordysphagia，andaprediction model P1 wasestablished.At the same time，the earlywarningmodel P2 of the improved data setwasonstructedasedontheMOEalgorittoexpandthlowsampledtasetndteprediciveailityofteodelwaseaatedbyte receiveroperatigueOCsulseasdeasdoesicsioslaedinsedfct were independent risk factors for dysphagia in patients with acute stroke（ P<0.05） .The prediction rateof model P1=1/[1+e-（-5.102+0.044×X1+0.164×X2+1.141×X3+0.725×X4）] ， （20 P2 =11 +e P2 model wassignificantly higherthan thatof P1 modelt=10.642， （204 P<0.001 ），andthe sensitivity and specificity of P2 model in predicting swallowing dysfunction were higher than those of P1 model.The calibration curveshowstatthemodelpredictshighconsistencybetweenthepredictedprobabilityandtheactualprobabiltofacutestrok.Cocusion Increasedgeesdsiladistdzoelyaedtalogui inpatientswithcuetroke.ouhteOEoversamplingagoi，teidividualiedalwaingodelasedonteaboeisctos ahigh diagnostic efficiency for swallowing dysfunction in patients with acute stroke.

KeyWords：Acute ischemic stroke;Swallowingdysfunction;Oversampling;Model

吞咽功能障碍（swallowingdysfunction）是脑卒中（stroke）后常见的并发症，患者主要表现为吞咽困难及饮水或进食呛咳等，不仅可导致卒中患者产生心理障碍，且由于进食困难，还可进一步引起患者饮食量减少，导致营养摄入不足，并增加了吸入性肺炎，甚至室息等并发症的发生风险。（剩余7668字）