Table 4.
Model performance measures.
| Clinical model | Multi-biomarker model | Difference | |
|---|---|---|---|
| All-cause readmission | |||
| Model’s log-likelihood | −1657.01 | −1611.83 | 45.18 |
| McFadden R2 | 0.08 | 0.11 | 0.03 |
| AIC | 3338.02 | 3253.66 | 84.36 |
| BIC | 3395.53 | 3325.55 | 69.98a |
| CV-related readmissions | |||
| Model’s log-likelihood | −1247.12 | −1208.16 | 38.96 |
| McFadden R2 | 0.08 | 0.11 | 0.03 |
| AIC | 2518.24 | 2446.32 | 71.92 |
| BIC | 2575.75 | 2518.21 | 57.54a |
| HF-related readmission | |||
| Model’s log-likelihood | −852.02 | −819.34 | 32.68 |
| McFadden R2 | 0.09 | 0.12 | 0.04 |
| AIC | 1726.04 | 1666.68 | 59.36 |
| BIC | 1778.76 | 1733.78 | 44.98a |
Multi-biomarker model – NT-proBNP, hs-TnT, and ST2 on top of clinical model.
Clinical model – see Section “Materials and Methods” for details.
aDifferences >10 are defined as very strong in favor of the model with lower values.