2024
HEART: Learning better representation of EHR data with a heterogeneous relation-aware transformer
Huang T, Rizvi S, Thakur R, Socrates V, Gupta M, van Dijk D, Taylor R, Ying R. HEART: Learning better representation of EHR data with a heterogeneous relation-aware transformer. Journal Of Biomedical Informatics 2024, 159: 104741. PMID: 39476994, DOI: 10.1016/j.jbi.2024.104741.Peer-Reviewed Original ResearchElectronic health recordsElectronic health record datasetDownstream tasksLanguage modelModeling electronic health recordsLearning better representationsPretrained language modelsEntity predictionRepresentation learningAnomaly detectionAttention weightsRelation embeddingsHealthcare applicationsEncoding schemeMed-BERTHigher-order representationsInput sequenceComputational costReadmission predictionPairwise relationshipsEHR dataElectronic health record dataSuperior performanceHeterogeneous contextsMedical entitiesIdentifying signs and symptoms of urinary tract infection from emergency department clinical notes using large language models
Iscoe M, Socrates V, Gilson A, Chi L, Li H, Huang T, Kearns T, Perkins R, Khandjian L, Taylor R. Identifying signs and symptoms of urinary tract infection from emergency department clinical notes using large language models. Academic Emergency Medicine 2024, 31: 599-610. PMID: 38567658, DOI: 10.1111/acem.14883.Peer-Reviewed Original ResearchElectronic health recordsNatural language processingNatural language processing modelsEmergency departmentTransformer-based modelsClinical notesF1-measureClinical decision supportLanguage modelSpaCy modelsU.S. health systemElements of natural language processingPublic health surveillanceConvolutional neural network-based modelProcessing long documentsIdentification of symptomsHealth recordsHealth systemClinician notesNeural network-based modelMedical careHealth surveillanceSymptom identificationEntity recognitionNetwork-based modelIdentifying incarceration status in the electronic health record using large language models in emergency department settings
Huang T, Socrates V, Gilson A, Safranek C, Chi L, Wang E, Puglisi L, Brandt C, Taylor R, Wang K. Identifying incarceration status in the electronic health record using large language models in emergency department settings. Journal Of Clinical And Translational Science 2024, 8: e53. PMID: 38544748, PMCID: PMC10966832, DOI: 10.1017/cts.2024.496.Peer-Reviewed Original ResearchElectronic health recordsNatural language processingHealth recordsIncarceration statusSignificant social determinant of healthSocial determinants of healthClinic electronic health recordsEHR databasePopulation health initiativesDeterminants of healthMitigate health disparitiesRacial health inequitiesEmergency department settingICD-10 codesHealth inequalitiesNatural language processing modelsHealth disparitiesHealth initiativesDepartment settingEmergency departmentSystem interventionsICD-10Clinical notesStudy populationLanguage model
2021
Characterizing physician EHR use with vendor derived data: a feasibility study and cross-sectional analysis
Melnick ER, Ong SY, Fong A, Socrates V, Ratwani RM, Nath B, Simonov M, Salgia A, Williams B, Marchalik D, Goldstein R, Sinsky CA. Characterizing physician EHR use with vendor derived data: a feasibility study and cross-sectional analysis. Journal Of The American Medical Informatics Association 2021, 28: 1383-1392. PMID: 33822970, PMCID: PMC8279798, DOI: 10.1093/jamia/ocab011.Peer-Reviewed Original ResearchConceptsElectronic health recordsEHR timeCross-sectional analysisAmbulatory physiciansPatient timeHealth systemClinical hoursHours of patientsMedStar Health systemYale-New HavenObstetrics/gynecologyNeurology/psychiatryMultivariable analysisPhysician genderCertain medical specialtiesPhysical medicineFemale physiciansEHR usePhysiciansHealth recordsHealthcare systemMedical specialtiesHoursSpecialtiesGender