Table 1. (A) A summary of existing prognostic systems. Many existing prognostic systems utilize a similar feature set, but differ considerably in their types (continuous vs categorical) and in the predictors constructed with those features in the original manuscripts. (B) A summary of the number of patients at each time threshold with patients not having complete data at each threshold being censored.
Prognostic Scoring System | Feature List | Feature type | Original Model/Predictor | |
---|---|---|---|---|
A: Summary of existing prognostic systems | ||||
SM Classification1 | Maximum nidus diameter | Categorical: 1–3 | Quadratic regression (∑ features) | |
Location (±critical function) | Binary: 0, 1 | |||
Venous drainage | Binary: 0, 1 | |||
RBAS1 | Volume | Continuous | Linear regression (∑ weighted features) | |
Location (deep vs superficial) | Binary: 0, 1 | |||
Age | Continuous | |||
VRAS1 | Volume | Categorical: 0–2 | Logistic regression (∑ features) | |
Location (±critical function) | Binary: 0, 1 | |||
History of hemorrhage | Binary: 0, 1 | |||
B: Number of patients at each time threshold | ||||
Time Threshold | Number of patients | |||
Year 2 | 1674 | |||
Year 3 | 1586 | |||
Year 4 | 1442 | |||
Year 5 | 1386 | |||
Year 6 | 1340 | |||
Year 7 | 1308 | |||
Year 8 | 1291 |
1SM = Spetzler-Martin; RBAS = modified radiosurgery-based AVM score; VRAS = Virginia radiosurgery AVM scale.