Journal: Heart Rhythm O2
Article Title: Validation, bias assessment, and optimization of the UNAFIED 2-year risk prediction model for undiagnosed atrial fibrillation using national electronic health data
doi: 10.1016/j.hroo.2024.09.010
Figure Lengend Snippet: Model validation and optimization flowchart from the original UNAFIED-10 model to the new UNAFIED-8 model. AF = atrial fibrillation/atrial flutter; BMI = body mass index; CDS = clinical decision support; COPD = chronic obstructive pulmonary disease; EHR = electronic health record; HF = heart failure; INPC HIE = Indiana Network for Patient Care health information exchange; OSA = obstructive sleep apnea; POC = proof-of-concept; UNAFIED = undiagnosed atrial fibrillation prediction using electronic health data. ∗Refer to Grout et al. 20 † Patient subpopulations were based on sex, type of health insurance, race, and region.
Article Snippet: The C-statistic (95% CI) was similar (0.850 [0.845–0.857] vs 0.851 [0.849– 0.851]) ( ) between the complete and balanced validation cohorts from Optum EHR discrete data, which was not improved with NLP data enrichment (0.847 vs 0.842) (Online ).
Techniques: Biomarker Discovery