Predictive model for in-hospital acute cerebral infarction in patients with acute gastrointestinal bleeding: a retrospective cohort study

Background Acute gastrointestinal bleeding (GIB) is a common emergency in gastroenterology and may be accompanied by blood loss, anemia, systemic inflammation, and hemodynamic instability. Because the brain has a high metabolic demand and depends on continuous perfusion and oxygen delivery, patients with GIB may be vulnerable to in-hospital acute cerebral infarction (ACI). However, whether GIB is associated with a higher incidence of in-hospital ACI in gastroenterology inpatients remains insufficiently studied, and clinically practical risk stratification tools are lacking. Objective To compare the incidence of in-hospital ACI between patients with and without GIB, identify factors associated with in-hospital ACI among patients with GIB, and develop a clinically applicable nomogram. Methods A retrospective cohort study was conducted among gastroenterology inpatients at Nanchong Central Hospital between September 2020 and December 2025. Patients were classified as GIB or non-GIB. Propensity score matching (PSM) was used to compare in-hospital ACI incidence between groups. Among patients with GIB, candidate predictors were evaluated using univariable and multivariable logistic regression, and a nomogram was developed and internally validated. Sensitivity, alternative, and exploratory subgroup analyses were additionally performed. Results A total of 2,734 patients with suspected GIB and 10,763 patients without GIB were initially screened; after exclusion, 2,380 and 8,996 patients, respectively, were included. After 1:1 PSM, 2,380 matched pairs were analyzed. In-hospital ACI occurred more frequently in the GIB group than in the non-GIB group (7.8% vs. 5.2%, p 0.001). Among patients with GIB, previous cerebral infarction history (OR 13.47, 95% CI 9.26–19.60), sepsis and infection (OR 2.43, 95% CI 1.71–3.44), cerebral hemorrhage (OR 3.48, 95% CI 1.17–10.33), anemia (OR 1.54, 95% CI 1.02–2.32), age (OR 1.05 per year, 95% CI 1.03–1.07), and length of hospital stay (OR 1.04 per day, 95% CI 1.02–1.07) were independently associated with in-hospital ACI. The nomogram demonstrated good discrimination (AUC 0.864, 95% CI 0.837–0.888) and calibration (MAE 0.008). Conclusion In this single-center retrospective cohort, GIB was associated with a higher incidence of in-hospital ACI. A six-factor nomogram based on routinely available clinical variables showed good internal performance and may assist risk stratification, although external validation is needed before broader clinical use.