Loading prediction engine

Connecting to prediction server...

0s, cold boot

About the giant aneurysm strategy model

Giant intracranial aneurysms often cannot be treated with a clip alone. This tool estimates the probability of a good functional outcome (modified Rankin Scale 0 to 3) under three microsurgical strategies: clip only, bypass with clipping, and bypass with parent artery occlusion.

Per-strategy probabilities are associational. A separate causal model estimates the effect of bypass versus clipping and is presented as hypothesis-generating. The tool also shows the most similar prior cases and their outcomes.

Predicts: Probability of mRS 0 to 3 for each strategy, and a causal bypass versus clip comparison.
Inputs: Age, sex, presentation (hemorrhagic or not), aneurysm site, multiple aneurysms and planned strategy.
Model: Random forest and k-nearest-neighbor ensemble, benchmarked against 13 model families.
Validation: Single-center series of 81 patients. Out-of-fold (repeated 5-fold and leave-one-out): AUC 0.715, Brier 0.184, calibration error 0.082.

Key terms

Research demonstration only. Not for clinical decision-making.