Additional file 1: of The prediction accuracy of dynamic mixed-effects models in clustered data

Abstract

Supplementary Material. Figure S1. Effect of the size of the random intercept and random slope on clinic-level clustering. Figure S2. Effect of the size of the residual error on the rate of improvement in prediction accuracy at a given clinic. Figure S3. Effect of the size of the residual error on model prediction accuracy. Figure S4. Effect of unknown patient-level predictor on the rate of improvement in prediction accuracy at a given clinic. Figure S5. Effect of the update interval on model prediction accuracy. Figure S6. Effect of the update interval on model prediction accuracy by clinic-size quintile. Figure S7. Relationship between bias in estimated model coefficients and prediction accuracy for the linear model. Figure S8. Relationship between bias in estimated model coefficients and prediction accuracy for the BLME model with random intercept. Figure S9. Relationship between bias in estimated model coefficients and prediction accuracy for the BLME model with random intercept and slope. Figure S10. Effect of extreme values of I˛ 2 on model prediction accuracy. Figure S11. Relationship between bias in estimated model coefficients and prediction accuracy for the BLME model with random intercept, with clinic size influencing the outcome. Figure S12. Relationship between bias in estimated model coefficients and prediction accuracy for the BLME model with random intercept and slope, with clinic size influencing the outcome. Figure S13. Effect of a non-linear relationship in the known patient-level predictor on model prediction accuracy. Figure S14. Effect of a non-linear relationship in the unknown patient-level predictor on model prediction accuracy. (PDF 3602 kb)