Journal: BMJ Health & Care Informatics

Article Title: Finding a constrained number of predictor phenotypes for multiple outcome prediction

doi: 10.1136/bmjhci-2024-101227

Figure Lengend Snippet: Task 1–3 results. The test set AUROC performance of the different models (constrained LR, constrained GBM, most-comprehensive-case LR, simplest-case LR and existing models) across the databases for the prediction tasks with existing models. The red lines are the 95% CIs. The Optum EHR and Optum CDM databases are shaded since these databases were not used to learn the predictor phenotypes. AUROC, area under the receiver operating characteristic curve; Optum CDM, Optum’s de-identified Clinformatics Data Mart Database; Optum EHR, Optum de-identified Electronic Health Record. GBM: gradient boosting machines; LR: logistic regression.

Article Snippet: Two held-out databases (Optum de-identified Electronic Health Record data set (Optum EHR) and Optum’s de-identified Clinformatics Data Mart Database (Optum CDM)) were only used to develop the models using the constrained predictors.

Techniques:

Journal: BMJ Health & Care Informatics

Article Title: Finding a constrained number of predictor phenotypes for multiple outcome prediction

doi: 10.1136/bmjhci-2024-101227

Figure Lengend Snippet: Task four results. The test set AUROC performance of the different models (constrained LR, constrained GBM, most-comprehensive-case LR and simplest-case LR) for the five outcomes across the databases. The red lines are the 95% CIs. The Optum EHR and Optum CDM databases are shaded since these databases were not used to learn the predictor phenotypes. AUROC, area under the receiver operating characteristic curve; Optum CDM, Optum’s de-identified Clinformatics Data Mart Database; Optum EHR, Optum de-identified Electronic Health Record. GBM: gradient boosting machines; LR: logistic regression.

Article Snippet: Two held-out databases (Optum de-identified Electronic Health Record data set (Optum EHR) and Optum’s de-identified Clinformatics Data Mart Database (Optum CDM)) were only used to develop the models using the constrained predictors.

Techniques:

Journal: Contemporary Clinical Trials Communications

Article Title: Using passive extraction of real-world data from eConsent, electronic patient reported outcomes (ePRO) and electronic health record (EHR) data loaded to an electronic data capture (EDC) system for a multi-center, prospective, observational study in diabetic patients

doi: 10.1016/j.conctc.2022.100920

Figure Lengend Snippet: Diagram of the end-to-end data flow CDR = Common Data Repository; EHR = electronic health record; ePRO = electronic patient-reported outcome; EDC: Electronic data capture; PI: Principal Investigator.

Article Snippet: Optum confirmed that all required study data was present and available in the EHR systems as described below. provides a pictorial view of the mapping process from EHR to study database.

Techniques: