Nature Communications · 2023 · 57 citations · 26 references
ComputingHealth InformaticsRadiologyDiagnostic ImagingElectronic Health RecordsMachine LearningComputer-aided DiagnosisDeep LearningDiabetes MellitusDisease ModelingElectronic Health Record AnalyticsMedical ImagingDeep Learning DetectionDeep Learning In RadiologyDeep Learning-based DetectionDeep Learning-based DiagnosisMachine Learning For HealthcareMachine Learning Prediction ModelsModel Predictivity AssessmentRadiologic Imaging
Deep learning (DL) models can harness electronic health records (EHRs) to predict diseases and extract radiologic findings for diagnosis. With ambulatory chest radiographs (CXRs) frequently ordered, we investigated detecting type 2 diabetes (T2D) by combining radiographic and EHR data using a DL model. Our model, developed from 271,065 CXRs and 160,244 patients, was tested on a prospective dataset of 9,943 CXRs. Here we show the model effectively detected T2D with a ROC AUC of 0.84 and a 16% prevalence. The algorithm flagged 1,381 cases (14%) as suspicious for T2D. External validation at a distinct institution yielded a ROC AUC of 0.77, with 5% of patients subsequently diagnosed with T2D. Explainable AI techniques revealed correlations between specific adiposity measures and high predictivity, suggesting CXRs' potential for enhanced T2D screening.
26
Index for rating diagnostic tests
W. J. Youden · Cancer · 1950 · 11.5K citations · Full text
Education, Educational Assessment, Educational Measurement +11
Economic Costs of Diabetes in the U.S. in 2017
Diabetes Care · 2018 · 2.6K citations · Full text
Danielle Butler, Stephen Petterson, Robert L. Phillips et al. · Health Services Research · 2012 · 671 citations
Public Affairs, Community Health, Health Services Research +16