AI Science Tracker
EN

Registry models link prescribing patterns to clinic and medication context

Stanford University; American College of Cardiology · 24 Nov 2022 · medicine

Discovery

Registry models link prescribing patterns to clinic and medication context

Source check: Source checked. Not independent scientific confirmation.

Finding: In 864,339 registry patients with atrial fibrillation, an enhanced XGBoost model identified documented anticoagulant prescriptions with a test AUROC of0.811. Clinic-location income, clinic size and region contributed permutation importance alongside age and concurrent medications. These are descriptive prescribing-pattern associations, not causal socioeconomic effects.

Technology: XGBoost; random forest; penalized logistic regression

Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

AI contribution
Fitted classifiers map patient and clinic features to documented prescriptions. Permutation analysis on the test set identifies model-used contextual information beyond a clinical risk-score feature set.
Test and result
PINNACLE records from January2017–June2018 were split80:20 by patient; hyperparameters used fivefold training cross-validation. Prescription status and inputs came from the most recent three-month record window.
Limitation / still unknown
Income is a clinic-ZIP-code mean, not each patient’s household income. Concurrent medications and INR can reflect prescribing or monitoring consequences rather than causes; same-clinic records can occur on both sides of the random split. Missing fields and voluntary outpatient practices limit representativeness. The CHA2DS2-VASc comparator is a clinical stroke-risk score, not a prescription model; no causal disparity, guideline benefit or improved treatment outcome is established.
Evidence sources and annotation context

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC10041076/ · https://www.ncbi.nlm.nih.gov/research/bionlp/RESTful/pmcoa.cgi/BioC_json/PMC10041076/unicode

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 21 Nov 2026. Review scope: Official primary BioC original article sections: Registry/outcome/feature Methods;patient split/tuning;test permutation-importance Results and limitations; no laboratory-method validation.

No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.

Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Open in the full catalog for filters, charts and related discoveries. Browse medicine discoveries.

Related catalog entries

Links use documented relationships where available, plus shared catalog context. A shared topic or model family does not establish a scientific dependency.