基于机器学习算法构建晚期直肠癌病人疼痛危象预测模型

打开文本图片集

AbstractObjective：Toconstructaprediction modelforpaincrisisinpatients withadvancedrectalancerbased onmachine learning algorithmsandanalyzethepredictiveperformanceofdiferentmodels.Methods：Aconveniencesamplingmetodwasusedtoselect21 patientswithadvancedrectalcaneradmitedtoourhosptalfromSeptember22toSeptember224asthestudysubjects.Questioires wereadministeredusingtheGeneralIformationQuestioaire，ocialSupportRatingScale，ConorDavidsonResiliencealeand HospitalAnxietyandDepressonScale.Basedonwhetherpatientsexperiencedpaincrisis，theyweredividedntoapaincrsisgroupand anon-paincrisisgroup.Univarateandmultivariateanalyseswereconductedtoidentifyinfluencingfactorsofpainrsis.Predictionmodels wereconstructedusingLogisticregressonandomforest，andecisiontrealgoritmsbasedontheuivariateandultivaateaalysis results.Thereceiveroperatingcharacteristic （ROC）curveandtheareaunder thecurve（AUC）wereused toevaluate modeleficacyand predictive value.Results：Among the 2lO patients with advanced rectal cancer，64 （30.48% ）experienced paincrisis.Multivariateanalysis showedthatsocialsupport，psychologicalesilienceegatieemotions，ag，monthlyouseholdicomepercapitandthebeof radiotherapy/chemotherapysessionswereindependentinfluencingfactorsforpaincrisisinpatientswithadvancedrectalancer（all P< 0.05）ROCralseadatthUClufgisicgessoelsitreldo modelwereO.90.90lnd.9，espctivelyeacuracyrateswere.881.852nd.889;sensivityrteswere5， and0.824;speifiyatesee938，.90，nd913;ecallteswere.534ad024;pcisioatesre04d 0.93;andFsdspieliteced AUC，accuacysitiallisiodFse，mosratigstrallpfane.Cocui：dfot modeldemonstratessuperiorpredictiveperformanceforpaincrises inadvancedrectalcncerpatientscompared toLogisticregesioand decisiontreodelsicalyisdelanelentifyigiskpatintsfoarlytevetoitpreventimeasues，tebui the incidence of pain crises.

Keywordsadvancedrectalancer;paincrisis;machineleaingalgorithms;predictionmodel;socialsupport;psyhologicalresilience摘要目的：基于机器学习算法构建晚期直肠癌病人疼痛危象的预测模型并分析不同模型的预测性能。（剩余10355字）