Construction of a nomogram model for predicting hematoma enlargement risk in hypertensive intracerebral hemorrhage:based on non-contrast CT features and clinical parameters
CUI Yanqiu
ZHENG Xueying
SHI Hao
ZHU Qi
Abstract：Objective To construct a prediction model based on non-contrast computed tomography(CT)features and clinical parameters for assessing the risk of hematoma enlargement in patients with hypertensive intracerebral hemorrhage(HICH)and to evaluate its predictive efficacy.Methods The clinical and imaging data of 216 HICH patients admitted to Suzhou First People's Hospital from December 2021 to May 2024 were retrospectively analyzed.Patients were categorized into a hematoma enlargement group(n=84)and a non-hematoma enlargement group(n=132)based on the criteria of hematoma enlargement exceeding 33%in relative volume or 6 mL in absolute volume.The dataset was randomly split into training and test sets in a 7:3 ratio.The candidate predictors were selected using univariate analysis,LASSO regression,and the Max-Relevance and Min-Redundancy(mRMR)algorithm.Multivariate logistic regression was employed to establish a nomogram model for predicting hematoma enlargement in HICH patients,as well as the related clinical and hematoma sign models.The predictive efficacy of each model was assessed using area under receiver operating characteristic(ROC)curve(AUC)and calibration curve,and clinical decision curve analysis was used to evaluate the clinical utility of the models.Patients were then categorized into low-and high-risk subgroups.Results The results of the CT scan review revealed that hematoma enlargement occurred in 84 patients,representing an incidence of 38.89%(84/216).Univariate analysis indicated statistically significant differences between the hematoma enlargement and non-hematoma enlargement groups in terms of age,interval between CT reexamination,irregular use of antihypertensive medication,pre-admission Glasgow Coma Scale(GCS)score,presence of cerebral hernia,ventricular involvement,hematoma volume,hematoma location,hemorrhage margin irregularity,as well as the swirl sign,blend sign,satellite sign,and island sign(P<0.05).Moreover,the nomogram model demonstrated the most favorable predictive efficacy(AUC=0.926)when compared to the clinical model and the hematoma sign model in the training set.Furthermore,the AUC value of the nomogram model in the test set reached 0.951.Calibration curves for the nomogram model in both the training and test sets showed satisfactory results,exhibiting good agreement(Hosmer-Lemeshow P>0.05).Clinical decision curve analysis revealed that the nomogram model provided superior clinical decision guidance for assessing hematoma enlargement risk in HICH patients.The incidence of hematoma enlargement differed significantly between the high-risk and low-risk subgroups(P<0.001).Conclusions This study constructs a nomogram model based on non-contrast CT features and clinical parameters,including hematoma volume,hemorrhage margin irregularity,blend sign,swirl sign,satellite sign,island sign,age,irregular use of antihypertensive medication,interval between CT reexamination,and pre-admission GCS score.This model provides an effective personalized risk assessment tool for predicting the occurrence of hematoma enlargement in HICH patients,allowing early identification of high-risk individuals,and demonstrating potential for clinical application.
Keywords：hypertensive intracerebral hemorrhagehematoma enlargementcomputed tomographyhematoma signspredictionnomogram model
Publication Date：2025-12-28
Online Publishing Date：2025-12-02(First online date of this platform, not the publication date of the document)
Pages：10( 102-111 )