Locked model · β-Thalassemia
Model card · bthal
one calibrated XGBoost per endpoint · synthetic-labeled cohort
Lock manifest
Fitted-model manifest
Research preview
version
—
lockedAt
—
featureCount
—
trainN / testN
— / —
sha256_manifest
—
OSF
—
Zenodo
—
recalibration
isotonic / Platt by ECE
Provenance
Model provenance · synthetic research-preview
Research preview
Model
—
Method
one calibrated XGBoost classifier per endpoint
Training data
Synthetic-labeled cohort (no PHI). Labels from a literature-signed ground truth pinned to published base rates — NOT real patient data.
External validation
None. Metrics are from a held-out synthetic test split.
Pre-registration
Not registered — synthetic research-preview demonstrator (OSF_ID: synthetic-not-registered).
Deposit
Not deposited (Zenodo_DOI: synthetic-not-deposited).
Manifest sha256
—
Per-endpoint models · calibrated XGBoost · held-out synthetic test split
Calibration plot
Predicted vs observed · benefit head
Research preview
Decision curve analysis
Net benefit vs threshold · benefit head
Research preview
Net benefit on the held-out synthetic test split (Vickers 2006).
Locked feature vector
0 features
Endpoint criteria
Locked definitions per probabilistic output
Research preview
loading endpoint definitions…
This interface presents outputs of a research-preview prediction model. It is not a regulated decision-support tool. Predictions come from one calibrated XGBoost classifier per endpoint, fitted on a synthetic-labeled cohort (no PHI; labels anchored to published base rates) and evaluated on a held-out synthetic test split. Not pre-registered; not externally validated.