本研究构建了基于ChatGPT与DeepSeek双模系统的急诊医学分层教学模型，结合建构主义学习理论与布鲁姆教育目标分类学，设计了学习者画像与自适应教学策略。系统采用混合云架构，ChatGPT负责通用教学与交互，DeepSeek用于复杂临床推理。USMLE模拟测试准确率达87.6%，医学推理准确率达93%。一项为期3个月的随机对照试验（N=64）显示，实验组在知识掌握度、临床推理能力和学习满意度方面分别提高27%、35%和16个百分点（94%vs78%）。结果表明，该双模系统在急诊医学教学中具有显著效果，为AI辅助医学教育提供了可行的模型框架。

人工智能；急诊医学；分层教学；个性化学习；双模系统；随机对照试验

[1]BASNAWIA,KOSHAKA.ApplicationofArtificialIntelligenceinAdvancedTrainingandEducationofEmergencyMedicineDoctors:ANarrativeReview[J].EmergencyCareandMedicine,2024,1(03):247-59.

[2]NATESANS,BAILITZJ,KINGA,etal.ClinicalTeaching:AnEvidence-basedGuidetoBestPracticesfromtheCouncilofEmergencyMedicineResidencyDirectors[J].WestJEmergMed,2020,21(04):985-98.

[3]MOELLB,SANDARONSSONF,AKBARS.MedicalreasoninginLLMs:anin-depthanalysisofDeepSeekR1[J].FrontArtifIntell,2025,8:1616145.

[4]DONGH,LIOJ,SHERERR,etal.SomeLearningTheoriesforMedicalEducators[J].MedSciEduc,2021,31(03):1157-72.

[5]TANS,MILLSG.DesigningChinesehospitalemergencydepartmentstoleverageartificialintelligence-asystematicliteraturereviewonthechallengesandopportunities[J].FrontMedTechnol,2024,6:1307625.

[6]ASTERA,LAUPICHLERMC,ROCKWELL-KOLLMANNT,etal.ChatGPTandOtherLargeLanguageModelsinMedicalEducation-ScopingLiteratureReview[J].MedSciEduc,2025,35(01):555-67.

[7]FEIGERLOVAE,HANIH,HOTHERSALL-DAVIESE.Asystematicreviewoftheimpactofartificialintelligenceoneducationaloutcomesinhealthprofessionseducation[J].BMCMedEduc,2025,25(01):129.

[8]MANSOORM,IBRAHIMA,HAMIDEA.PerformanceofDeepSeekandGPTModelsonPediatricBoardPreparationQuestions:ComparativeEvaluation[J].JMIRAI,2025,4:e76056.

[9]SALMANH,POWELLL,ALSUWAIDIL,etal.ProfilingofLearnersinMedicalSchoolsasaMoveTowardPrecisionEducation:ProtocolforaScopingReview[J].JMIRResProtoc,2022,11(10):e41828.

[10]杨海燕,李涛.ChatGPT教学应用:场景、局限与突破策略[J].中国教育信息化,2023,29(6):26-34.

[11]SOENKSENLR,MAY,ZENGC,etal.Integratedmultimodalartificialintelligenceframeworkforhealthcareapplications[J].NPJDigitMed,2022,5(01):149.

[12]TEMSAHA,ALHASANK,ALTAMIMII,etal.DeepSeekinHealthcare:RevealingOpportunitiesandSteeringChallengesofaNewOpen-SourceArtificialIntelligenceFrontier[J].Cureus,2025,17(02):e79221.

[13]DAVENPORTT,KALAKOTAR.Thepotentialforartificialintelligenceinhealthcare[J].FutureHealthcJ,2019,6(02):94-8.

[14]PETERSSONL,LARSSONI,NYGRENJM,etal.Challengestoimplementingartificialintelligenceinhealthcare:aqualitativeinterviewstudywithhealthcareleadersinSweden[J].BMCHealthServRes,2022,22(01):850.

[15]VANLANKVELDT,SCHOONENBOOMJ,KUSURKARRA,etal.Integratingtheteachingroleintoone'sidentity:aqualitativestudyofbeginningundergraduatemedicalteachers[J].AdvHealthSciEducTheoryPract,2017,22(03):601-22.

[16]RINCONEHH,JIMENEZD,AGUILARLAC,etal.Mappingtheuseofartificialintelligenceinmedicaleducation:ascopingreview[J].BMCMedEduc,2025,25(01):526.