ABSTRACT NUMBER: ESOC2026A1732 DEVELOPMENT AND INTERNAL VALIDATION OF A CLINICALLY INTERPRETABLE NOMOGRAM TO PREDICT DISCHARGE FUNCTIONAL OUTCOME IN CEREBRAL AMYLOID ANGIOPATHY–RELATED LOBAR INTRACEREBRAL HEMORRHAGE

Abstract Background and aims Prognostic evidence specific to cerebral amyloid angiopathy (CAA) related lobar intracerebral hemorrhage (ICH) remains limited, our study aimed to develop a short-term prognostic model to provide evidence for long-term outcome prediction and to inform clinical care planning and resource allocation. Methods We retrospectively enrolled 124 patients aged ≥50 years with CAA-related lobar ICH diagnosed based on the Boston criteria version 2.0 and treated at Beijing Tiantan Hospital from 2019 to 2024. Discharge modified Rankin Scale (mRS) was dichotomized as favorable (0–2) versus unfavorable (3-6). Candidate predictors were screened using the Boruta algorithm, followed by multivariable logistic regression with backward selection and multicollinearity assessment. A nomogram was constructed from the final predictors, and performance was evaluated by discrimination, calibration, bootstrap internal validation (500 resamples), and decision-analytic (DCA) measures. Results Of 124 patients, 54 (43.5%) had favorable outcomes and 70 (56.5%) had unfavorable outcomes at discharge. The final model included hematoma volume, glucose, prior cognitive decline, cSS grade, admission NIHSS, and subarachnoid hemorrhage. The nomogram demonstrated good discrimination (AUC 0.917) and calibration, with bootstrap validation supporting robustness; DCA suggested net benefit across clinically relevant thresholds. Conclusions A six-variable, clinically interpretable nomogram provides accurate early risk stratification for discharge functional outcomes in CAA-related lobar ICH and may support individualized in-hospital management and discharge planning. Conflict of interest Yingjie Zhang and Ju Yi: no conflict of interest requiring disclosure.