External Validation Results for Stroke Medication Adherence Models

External validation results evaluate beta regression models predicting 1-year statin and antiplatelet adherence in post-stroke patients. The dataset includes performance metrics for two cohorts: 2369 patients for statins and 2147 for antiplatelets. It contains model assessment statistics like R², mean prediction difference, and calibration slope.

Use Cases

• Compare the R² values (0.67 for statin, 0.56 for antiplatelet) to benchmark new adherence prediction models.
• Analyze the difference between mean observed and mean predicted Proportion of Days Covered (PDC) to assess model bias.
• Evaluate calibration slope results (1.06 and 0.96) for model reliability across the statin and antiplatelet cohorts.
• Use the cohort sizes (2369 and 2147 patients) to inform sample size calculations for similar validation studies.

Strengths

• Contains validation results for two distinct patient cohorts with specific sizes: 2369 for statins and 2147 for antiplatelets.
• Reports concrete model performance metrics: R², mean PDC difference, and calibration slope.
• Focuses on a clinically significant outcome: predicting 1-year medication adherence after stroke.

Limitations

• Very small scale at 5.5 KB, indicating the dataset likely contains summary results, not the underlying patient-level data.
• Lacks the raw patient data or feature columns needed to retrain or deeply analyze the models.
• Temporal coverage and specific geographic origin of the validation data are not provided.

Provenance

Source

Elias Edward Tannous via figshare.

Collection Method

Model performance was assessed using external validation on prescription-filling data to calculate Proportion of Days Covered (PDC).

Time Range

null

Freshness

Last updated in March 2026.

Geography

null