Risk Factors and Model Construction for Early Neurological Deterioration in Patients with Intracerebral Hemorrhage

Heng ZhouDapeng DaiKang Xie^*Aimin Li^*

• The First People’s Hospital of Lianyungang, Lianyungang, China

Objective: To investigate the risk factors for early neurological deterioration (END) in patients with spontaneous intracerebral hemorrhage (ICH), construct a predictive model, and evaluate its predictive efficacy. Methods: We retrospectively selected 450 ICH patients admitted to the First People’s Hospital of Lianyungang from June 2023 to September 2024. The patients were randomly divided into a training set (315 patients) and a validation set (135 patients) at a 7:3 ratio. In the training set, patients were categorized into END group (n=66) and non-END group (n=249) based on the criteria of a decrease in GCS score by ≥2 points or an increase in NIHSS score by ≥4 points within 72 hours of admission. We compared the general data, laboratory test results, and imaging features between the two groups. We used LASSO regression and multivariate logistic regression analysis to identify the independent risk factors for END in ICH patients. A nomogram model for predicting END in ICH patients was constructed using the R language rms package and applied to the validation set to assess the model’s predictive ability and accuracy by drawing ROC curves, calibration curves, and decision curve analysis (DCA) curves. Results: In the training set, there were significant differences between the END and non-END groups in terms of age, admission systolic blood pressure, admission GCS score, admission NIHSS score, serum potassium, serum calcium, blood glucose, homocysteine (Hcy), white blood cell count (WBC), C-reactive protein (CRP), neutrophil-to-lymphocyte ratio (NLR), intraventricular hemorrhage (IVH), blend sign, midline shift, hematoma expansion (HE), and initial hematoma volume (P<0.05). The results of the LASSO regression and multivariate logistic regression analysis showed that the independent risk factors for END in ICH patients included age, WBC, Hcy, HE, blend sign, and admission systolic blood pressure. The area under the ROC curve (AUC) for predicting END in the training and validation sets were 0.909 and 0.831, respectively. The Hosmer-Lemeshow goodness-of-fit test showed that the model had good calibration (P=0.550 for the training set and P=0.368 for the validation set). Conclusion: .The nomogram model established based on these parameters can effectively predict END and provide a reference for clinical decision-making.