To explore the predictive value of endoscopic ulcer characteristics combined with serum albumin (ALB) levels for the prognosis of patients with peptic ulcer bleeding (PUB).

A retrospective study was conducted on 210 PUB patients admitted to Xingtai Central Hospital from May 2023 to May 2025, all of whom underwent emergency endoscopic examination and endoscopic hemostasis treatment. According to the patient's prognosis within 30 days, they were divided into a poor prognosis group (n=54) and a good prognosis group (n=156). Clinical data, endoscopic ulcer characteristics (including ulcer size, location, and Forrest classification), serum ALB, and blood urea nitrogen-to-creatinine ratio (BUN/Cr) were compared between the two groups. Logistic regression analysis was used to identify influencing factors for poor prognosis in PUB patients, and a predictive model was constructed. Receiver operating characteristic (ROC) curves were used to evaluate the predictive value of each independent factor and the combined prediction model.

There were statistically significant differences in SI, Forrest grading, maximum ulcer diameter, ALB, and BUN/Cr between the two groups (P<0.05). High SI, Forrest classification as high-risk, ulcer diameter >2 cm, low ALB, and high BUN/Cr were all risk factors affecting poor prognosis in PUB patients (P<0.05). The ROC curve analysis results showed that the area under the curve (AUC) of SI, Forrest grading, maximum ulcer diameter, ALB, BUN/Cr, and the nomogram prediction model were 0.756, 0.671, 0.685, 0.707, 0.796, and 0.941, respectively. Delong analysis revealed that the nomogram prediction model had the highest AUC and the best prediction performance (P<0.05). At the same time, the sensitivity and specificity of nomogram prediction model were the highest, which were 85.2% and 87.8% respectively.

The nomogram prediction model constructed in this study, which integrates endoscopic local features (Forrest high-risk grading, ulcer diameter >2 cm) and systemic serum biomarker (ALB), and incorporates SI and BUN/Cr, can effectively identify high-risk patients with poor prognosis after PUB. This model integrates the "local-systemic" risk dimension, and its predictive performance is significantly better than traditional single-index or scoring systems that rely solely on clinical parameters, providing a new approach for early stratified intervention in clinical practice.