Clinical Evidence Calibration in Acute Ischemic Stroke: Derived Evidence Profiles and Independent Replication Outputs

<p dir="ltr">This dataset contains derived analytical outputs from the development and independent replication of Clinical Evidence Calibration (CEC), an interpretable statistical framework for integrating heterogeneous clinical findings on a common evidence scale.</p><p dir="ltr">CEC represents individual clinical variables as signed evidence contributions and distinguishes unique evidence, shared or overlapping evidence, synergistic interactions, and conflicting evidence.

The dataset contains patient-level derived CEC scores, evidence-conflict measures, model coefficients, shared-evidence summaries, and interaction-selection stability obtained using nested cross-validation.</p><p dir="ltr">The framework was evaluated using three independently available acute ischemic stroke datasets comprising 3,073 analyzed records in total. The development analysis used a MIMIC-IV-derived acute ischemic stroke dataset (Zenodo DOI: 10.5281/zenodo.13980269).