Categorical Stroke Prediction Maintains Accuracy via Clinical Thresholds

Replacing continuous predictors with guideline-aligned categorical encodings for ischaemic stroke prediction shows negligible performance loss in most treatment cohorts. This discretization helps align machine learning model outputs with clinician reasoning and existing medical guidelines.