Loading prediction engine

Connecting to prediction server...

0s, cold boot

About the moyamoya ischemic-risk model

The Bypass Angiographic Scoring System (BASS) estimates the probability of a postoperative ischemic event after direct cerebral bypass for moyamoya disease. Inputs are preoperative segmental angiographic status (ICA, M1, A1, P2), donor-STA quality indicators, stoma and donor-recipient ratios, and the POD-1 direct-bypass supply territory.

Each prediction comes with a per-feature attribution: for every input, the predicted ischemic risk is recomputed after swapping that feature to the cohort median, and the shift (in percentage points) is shown so you can see which inputs are pushing the risk up or down for this patient.

Predicts: Probability of a postoperative ischemic event after direct cerebral bypass for moyamoya.
Inputs: Bypass side, donor count, segmental vessel status (ICA, M1, A1, P2), STA diameter ratio, STA kinking, stoma-to-recipient ratio, donor-to-recipient ratio and POD-1 bypass supply territory.
Outputs: Ischemic-event probability, risk band and a per-input attribution chart against the cohort baseline.
Model and validation: Deployed model is a calibrated SVM (BASS ischemic head). Reference WPFS neural model gives 20-repeat five-fold mean ROC-AUC 0.754 (SD 0.057), mean average precision 0.591 and mean balanced accuracy 0.699, on 38 anastomosis rows from 26 hemispheres and 21 patients.

Exploratory single-center cohort of 38 anastomosis rows. Related rows may cross training and validation splits; do not deploy as a validated clinical risk tool. Research demonstration only. Not for clinical decision-making.